Human Factors in Telecommunications? Links in Norway, Steinar M

Home , Misdialed call, Open communication, Tandberg, Telephonoscope

Contents

Feature Special

Guest editorial, Knut Nordby ...... 1 First installations of WDM, optical ADM and optical in-line amplifiers on long-haul cable Do we need human factors in telecommunications? links in Norway, Steinar M. Svendsen ...... 115 Knut Nordby ...... 2

Consumers, not Clients – Empowering users of new technology, Frode Volden ...... 9

Speech quality, usability and service quality, Norman Gleiss ...... 14 Status

Some current issues of Human Factors in International research and standardization telecommunications, Knut Nordby ...... 33 activities in telecommunication: Introduction, Per Hjalmar Lehne ...... 125 Feedback tones in public telephone networks: Human Factors work at ETSI, Donald M. Anderson .65 The ACTS Project AC215 CRABS; Cellular Radio Access for Broadband Services, The ergonomics of teleworking, John W. Bakke ...... 78 Agne Nordbotten ...... 126

Human factors of videotelephony, Knut Nordby ...... 90 AC066 OPEN – Open Pan-European Network, Torodd Olsen ...... 129 A tactile marker for telephone cards: – From idea to global standard, Knut Nordby ...... 98 MoMuSys, Robert Danielsen ...... 133

The User-Interface Design process; The SINUS project, Satellite communication some new year resolutions, Simon Clatworthy ...... 107 in the UMTS, Vendela Paxal ...... 139

Editorial office: Telektronikk Telektronikk Telenor AS, Telenor Research & Development P.O. Box 83 Volume 93 No. 3/4 - 1997 N-2007 Kjeller, Norway ISSN 0085-7130 email: [email protected] Editor: Editorial board: Ola Espvik Ole P Håkonsen, Senior Executive Vice President Tel. + 47 63 84 88 83 Oddvar Hesjedal, Vice President, Research & Development Bjørn Løken, Director Status section editor: Per Hjalmar Lehne Graphic design: Tel. + 47 63 84 88 26 Design Consult AS Editorial assistant: Layout and illustrations: Gunhild Luke Gunhild Luke, Britt Kjus, Åse Aardal Tel. + 47 63 84 86 52 Telenor Research & Development Guest editorial KNUT NORDBY

The possible services that can be information society. Again, the offered by today’s telecommuni- US is leading. In 1990 Congress cation systems are staggering, and passed the Americans with Dis- new services are continually being abilities Act (ADA), and a new added. However, developers and Telecommunications Act is now operators often seem to forget that going through Congress. This leg- the end users of these sophisti- islation provides for extensive cated services belong to the rights for the users and is being species Homo sapiens, which has closely studied by other countries. not changed significantly during In Europe, there is an increasing the last 50 000 years. Although demand for Pan-European legis- today’s computer-controlled lation and a mounting pressure on telecommunication networks the Commission of the European should be able to offer services Communities (CEC) to issue fully adapted to the capabilities of Directives to achieve the same human end users better than any level of consumer protection that previous technology, the opposite US citizens enjoy. is usually the case and new services that are not specifically In the days of the state-controlled adapted to human capabilities are monopolies, the consumers had no regularly introduced. Thus, the way to influence the quality and user interface of the plain old tele- usability of services and equip- phone (POT) has actually deterio- ment. Today, consumers’ and end- rated from the time of Alexander users’ organisations participate in Graham Bell to some of today’s standardisation, thus filling a void interactive voice services. In those left by telecom operators’ who are early days users interacted with other people and any misunder- reluctant to partake in standardisation – and these consumers’ standings could be set straight right away using the most basic of organisations may soon be the ones who are calling the shots. all human capabilities – the spoken language. Actually, the service providers and the consumers have the same objectives – good usability – but they may have different If people cannot easily use technology we will increasingly see solutions as how to achieve them. that the people in the Information Society will be divided into the ‘information rich’ and the ‘information poor’. The former For various reasons industry, and the telecom industry is no group will successively improve its position in society, while exception, has come to regard good usability or the misused the latter may be excluded from important societal services and term ‘user friendliness’ as costly add-ons that offer little become increasingly dependent on others. More than 25 % of advantage in the long run. Today, this view is being turned all European citizens are either aged, disabled or both, and these on its head. Simple cost/benefit analyses will show even the do not in any way make up any marginal groups. most stubborn executives the great economical and commercial advantage of good usability: When NYNEX, the great US East- During the last ten years a wave of liberalisation, de-regulation coast based telephone company, managed to shave just three and privatisation of the state-owned telecommunication monop- seconds average off each inquiry transaction, they saved them- olies has swept across Europe and other parts of the world. selves hiring more than 500 operators. These liberalised, privatised telecommunication markets need new regulation, and a wave of re-regulation and legislation is The bottom line of all this is Human Factors or Ergonomics, the now under way: the most liberalised market of them all, the US science of designing good user interfaces. The following nine market, is also the most regulated market to secure basic articles in this Feature Section of Telektronikk are devoted to services, just trading and fair competition. various perspectives of Human Factors in Telecommunications. These range from simple tactile identifiers on telephone cards to It is to these re-regulated markets that user-groups, special empowering users of new technology; from sound quality in the interest groups and consumers’ organisations are moving in, telephone to the special requirements of tele-working; from the demanding proper services and equipment. The trade journals basic tones (line signals) we hear in the handset to the intri- now print comparative tests of similar products, and these tests cacies of using video-telephones, and from the historic origins review usability; i.e. how easy is it to use, how long does it take of human factors in telephony (did Alexander Graham Bell to learn to use it, is it easy to make a fatal mistake. And think about it?) to a review of contemporary issues. A brief purchasing decisions are increasingly made on the basis of such overview like this can only scratch the surface of a large field, reviews. Prospective customers do not only look at prices – ease but hopefully this may inspire some readers to dip deeper into of use has become an important decision factor – a product with this fascinating topic. a bad usability mark is not going to make it in the marketplace. Thus, manufacturers and service providers are going to need all the human factors expertise they can get to keep a competitive advantage.

On the other side we have the legislators. Society cannot sit still and watch large groups of citizens being left outside the

Telektronikk 3/4.1997 1 Do we need human factors in telecommunications? KNUT NORDBY

NORWEGIAN TELECOMMUNICATIONS MUSEUM NORWEGIAN TELECOMMUNICATIONS MUSEUM The changing user interface

When the telephone was in its infancy, the user interface was the simplest of all times: You just lifted the handset off- hook and gave a small crank a turn or two to alert the other party, and when the call was answered you could start the conversation. This was in the days when only two and two telephones were connected together (e.g. between the home and the shop, or between the mansion Figure 1 Norwegian telephone set from Figure 3 Norwegian telephone set from house and the porter’s lodge). However, 1893 with inductor crank. This telephone 1931 with rotating number dial. This was the idea of connecting several telephones set worked with the simple user interface the first ‘modern’ telephone with inte- together via a switchboard or exchange from the early days of telephony. The grated hook switch. It was designed by was soon realised, making it possible to earpiece on the handset can be moved to Johan Christian Bjerknes of Elektrisk reach many people, thus making the tele- adapt to ‘small and large heads’, Bureau, Norway (with some finishing phone a much more important communi- a simple ergonomic consession to user touches by the artist Jean Heiberg), but it cation instrument. You still only needed variability was to be produced in Sweden by L.M. to lift the handset off hook and give the Ericsson, who had acquired a majority of little crank a turn or two to alert the oper- Elektrisk Bureau shares ator (in some systems you had to press the hook a couple of times), and when the exchange answered you simply asked As the telephone networks grew, opera- the telephone book had arrived. In the the operator to connect you to the party tors could no longer keep track of the beginning, this was not a serious problem you wished to speak to. When the call names and numbers of all subscribers, for most users, since the subscriber lists was finished, you merely replaced the and the telephone companies started were relatively short, phone numbers had handset on-hook (in some systems you issuing printed directories listing their only three or four digits and the print was had to give the crank another turn to alert subscribers’ names and numbers. Before large and legible. As the number of sub- the operator to disconnect you) – in other making a call you now had to look up the scribers increased, phone books grew words, a simple, intuitive, speech based number of the party you wished to speak bigger and heavier, while the print was user interface which could be used by to in the directory and communicate it to made smaller on low quality paper to anyone who could speak the language the operator – the telephone number and limit the size of the telephone book to (see Figures 1 and 2). keep costs down. This soon created problems, especially for elderly people with failing eye-sight and for visually NORWEGIAN TELECOMMUNICATIONS MUSEUM impaired people – a problem that persists to this day [1,2].

This was only the first of many turns for the worse in the telephone user interface. It was, of course, still possible to ask the operator to look up an unknown number, but this practice had to be dropped as traffic grew and the work load on operators increased. Instead directory services were introduced: Special numbers were established that you could call to access directory information. This information was, and still is, given by operators, with whom you can talk and solve misunder- standings. Initially, this service was free, but, eventually, it was turned into a revenue maker. For people with failing eye- sight or who are blind and who cannot look up numbers in a directory this has now become an economic problem since the prices for directory services have gone up, and alternatives (directories on Figure 2 The work of the telephone exchange operators (nearly always women) was a CD-ROMs or diskettes) are very expen- very demanding chore. The picture shows the main exchange of the Christiania sive or not available. In some countries Telephone Company in Christiania (now Oslo) in 1896

2 Telektronikk 3/4.1997 KNUT NORDBY

(e.g. Norway) blind subscribers are given cess you are, can be a daunting task even a limited number of directory enquiries for resourceful people, but may be impos- free, but these have to be made from the sible to perform by people with learning home (or work) telephone. However, it is disabilities or cognitive impairment. usually when you are on the move when Thus, many elderly or mentally retarded you have no directory available that you people may have to receive help from need the service most. another person to make a telephone call [1,2]. The introduction of the automatic telephone exchange and telephones with At first, direct dialling was only available turning number dials was a momentous for local calls, while long distance and setback for the simple user interface. In international calls still had to be handled addition to having to look up the number by operators. But by leaps and bounds in the directory, you now also had to be direct dialling ‘subscriber trunk dialling’ able to listen for, discriminate between has been extended, first for long distance and understand several line signals (the domestic calls, and eventually also for Figure 4 British (General Post Office) push key number various tones you hear in the handset, international calls. It is possible today to dialler. This device was used in exchanges to alleviate the such as dialling tone, ringing tone, en- call virtually any telephone in the world physical strain on operators who had to dial telephone gaged tone). But first and foremost, you directly from your own telephone with- numbers all day on a rotating dial had to ‘send’ the telephone number to the out the intervention of an operator, but exchange yourself. This was achieved by the price for this freedom (in addition to inserting the index finger in the hole at the cost of the call) is to figure out the the desired digit in the telephone dial and different call set-up procedures encoun- not always followed, often with serious turn the dial clockwise to the stop, with- tered, often necessitating the entering of consequences for the users – as demon- draw the finger to let the dial run back by long digit strings and codes (which today strated in the case of layout of digits and itself, and then repeat this procedure until may run into 11–12 digits) and the need symbols on telephone key-pads in Nor- all the digits in the telephone number had to be able to discriminate between and way. According to ITU-T Recommenda- been dialled. The movements required to understand many strange and unknown tion E.161 [10] the layout of digits and operate a turning telephone dial are hard line signals or tones – at the last count symbols on telephone dialling key-pads enough for young healthy people, but for Europe alone had 228 different line sig- shall be (as shown in separate box, right elderly people with rheumatism or arthri- nals! [3,4,5,6]. panel) with 1 2 3 on the upper row and * tis, or for people with upper limb motor 0 # at the bottom row. Norway, however, and dexterity impairments, operating the The introduction of tone signalling when tone signalling and push-button turning dial can be a serious or even (DTMF = Dual Tone Modulation Fre- telephones were introduced in 1981, insurmountable obstacle (see Figure 3). quencies), touch-tone to some, and the adopted a layout based on the Interna- replacement of the rotating dial with the tional Organization for Standardisation, For operators at telephone exchanges, the now ubiquitous twelve-key numeric ISO, 9995-4 [11] for the layout of digits physical strain of dialling telephone push-button key-pad has been a signifi- on calculator key-pads, with 7 8 9 at the numbers all day was so great that various cant step forward for all users: To press top and 0* #at the bottom (as shown in devices had to be introduced to lessen keys marked with digits are both physi- separate box, middle and right panels). their nasty chore. One such device (see cally easier and cognitively simpler than This was a well intended, but misguided, Figure 4) was a push-key number sender, turning a dial (see Figure 5). a device with ten typewriter-like push- keys numbered 1 through 9 + 0, to ease The advent of interactive stored voice NORWEGIAN TELECOMMUNICATIONS MUSEUM the heavy load imposed on index finger, supplementary services, however, in- hand, wrist and arm; anticipating the volving the use of the new * and # keys push-button telephone by several (referred to as star or asterisk, and decades. However, push-key diallers square, hash [UK] or pound [US], were normally not available to the ordi- respectively), have left many users dumb- nary subscriber, who would still have to founded. There is an urgent need for struggle with the turning dial for many standardisation and guidelines on the use years. This should, by the way, remind us of these two symbols as there are no that good user interfaces and improved standardised rules or recommendations usability is not only an issue for the sub- for their use today, and neither ITU-T scriber or end user, but is just as impor- (International Telecommunication Union, tant for all people working to provide the founded more than 125 years ago) nor service. Substantial savings have been ETSI (European Telecommunication made by telephone companies by im- Standards Institute, founded ten years Figure 5 ‘Tastafon’, the first Norwegian proving system interfaces and the ago) have made any serious attempts to ‘touch-tone’ push-button (DTMF) tele- operators’ work environment. rectify this situation by issuing harmonised phone from 1981. Push buttons are much recommendations or standards [7,8,9]. easier to operate than a rotating dial. The cognitive problems of setting up Note the inverted layout of the number calls, with no feedback other than the line Even when standards and recommenda- keys which does not follow ITU-T Rec- signal tones to tell you where in the pro- tions are available, however, these are ommendation E.161

Telektronikk 3/4.1997 3 Three different digit layouts of push button keypads used in the ‘Norwegian’ type layout (centre) based on the calculator Norway: Left, the ISO/IEC 9995-4 standard layout used on type layout (left) “to aid the users by only having to interact calculators and PC keyboards [11]. Centre, the ‘Norwegian’ with one common digit layout on telephones and calcula- layout, based on the calculator layout. Right, the interna- tors”. Today, both the ‘Norwegian’ and the ITU-T layouts are tional ITU-T, Recommendation E.161 standard layout of tele- used concurrently on telephones in Norway, making it nearly phone dialling keys [10]. When the push button ‘touch-tone’ impossible for blind and visually impaired people to dial num- telephone was being developed in Norway, early prototypes bers correctly when encountering a strange telephone, and carried the international CCITT (now ITU-T) standard digit creating a general nuisance to all users who have to switch layout (right). However, when the push button ‘Tastafon’ tele- between the two layouts [10,11]. phones were introduced in 1981, this had been changed to

ISO/IEC 9995-4 The ITU-T "Norwegian" Recommen- C = / x layout dation E.161

7 8 9 - 7 8 9 1 2 3

4 5 6 + 4 5 6 4 5 6

1 2 3 1 2 3 7 8 9 = 0 , 0 * # * 0 #

attempt to aid users by making telephone Telephone numbers are increasingly bloc dialling, where the number is first and calculator keypads similar, but which growing longer – many countries have composed on a display in the telephone has turned out to be an encumbrance to already adopted nine-digit internal tele- and then sent as a whole block to the ex- nearly all users now that telephones with phone numbers (e.g. France and United change by activating a “Send” button, (as ITU type key-pad layouts can be sold Kingdom) and there are plans for a Pan- with GSM and ISDN phones today), will alongside telephones with calculator type European numbering system running into have to replace traditional overlap key-pad layouts. Both visitors to Norway 13 digits, including the new four-digit dialling (where each digit is sent on to and Norwegians travelling abroad operator codes (e.g. 1531 for Telenor). the exchange as it is dialled). encounter problems when switching be- Many of the new interactive stored voice tween layouts, but those who are most services (e.g. tele-banking, tele-shop- affected are blind and visually impaired ping), and UPT (Universal Personal Human Factors in people. In Norway they have no way of Telephone, called Alpha number in Nor- telecommunications knowing which way the digits are ar- way) require entering long digit strings, ranged when encountering strange tele- including service codes, account numbers It was inevitable that the need arose for phones they have to use [12]. and PIN-codes (Personal Identification specialists to participate in the design of Number) – a development which Donald improved user interfaces. In 1947 Bell The concurrent use of different digit lay- Norman has referred to as “The conspira- Labs hired psychologist John Karlin to outs on telephones was not new to Nor- cy against the human memory” [13,14]. do exactly this, and the discipline human wegians, who for many years had to cope factors in telecommunications was born. with two different telephone dial layouts System feedback to users of Audiotex In the 50 years which have passed, the in Norway and a third layout in neigh- services is usually limited to a mindless number of human factors specialists, bouring Sweden (se separate box). ‘beep’ or at best to a courteous, but un- most of them psychologists, have risen to essential, voice message such as the ob- over 200 at AT&T alone, and most lead- Putting a tactile marker (e.g. a raised dot ligatory “Thank you for your order” to ing telecommunication research laborato- or dimple, which can be felt by the sense indicate that a valid action has been ries around the world have seen similar of touch) on the 5 key does not help since made, but feedback is seldom given to development. Human factors as such did the 4 5 6 keys are at the same positions confirm if the desired action has been not arise in 1947, but some years earlier. (i.e. the second row) in both key layouts correctly made [7,8,9]. Actually, its roots go back a good num- (also the # key has the same position in ber of years to the emergence of indus- both layouts). Initially, Denmark also The users must therefore devise their trial hygiene and industrial safety, and to used the calculator type layout on tele- own procedures and strategies to be able ergonomics based on antropometric phones, but soon realised the mistake and to remember and dial the long digit measurements (the measurement of man changed to the ITU type layout after only strings required, such as writing down and human abilities). a few years. People in both countries, numbers, splitting long numbers into however, will be hampered by the two groups (chunks), or using pre-pro- The concept of human factors (mainly concurrent layouts for many years to grammed automatic dialling devices. To US) or ergonomics (mainly UK) as a dis- come since telephones seem to be rather avoid the many mistakes users invariably tinct discipline did not arise within long-lived instruments [10,11,12]. will make as numbers grow longer, en- telecommunications, but with the mili-

4 Telektronikk 3/4.1997 MIKAEL JONSSON/MIRA/SAMFOTO

Three telephone dial layouts As a result of ordering equipment from producers tary during World War II when it became who followed different technical standards, the digits evident that the physical, cognitive and on telephone dials in Oslo and its environs (centre) mental abilities and the reactions of were laid out in the opposite order than on telephone human operators in stressed and unexpected situations had to be taken into dials in the rest of Norway (top), which comply to account when designing equipment and ITU Recommendation E.161 [10]. This created a systems. It got an impetus after a number great problem for blind people and others when of freak accidents as pilots, having flown travelling between Oslo and the rest of the country. dangerous missions and returned un- The close neighbourhood with Sweden created an harmed to their bases, unexplainably additional problem since Sweden used yet another crashed their planes after making some non-standard layout (bottom). However, blind users incomprehensible action, such as opening were not completely at a loss since they always the bomb bay instead of extending the could rely on geography to guide them as to which landing gear or similar slips. From its dial layout to expect. With the two numeric key-pad military uses, human factors thinking was taken up by various civilian fields layouts now in concurrent use this is not possible as such as civil aviation, industrial safety there is no relation between layout and geography; it and eventually, telecommunications. At is quite possible to encounter telephones with both first, the interests in telecommunications layouts even on the same desk. were mostly concerned with speech quality and traditional telephony, but has Figure 6 The success of the mobile tele- now moved into virtually every field of phone has been substantial and has telecommunications and information spread to all sectors of society (here we 4 3 technology (IT), from numbering plans see a Lapp user) despite the rather com- 5 2 to supplementary services, from Internet plicated user interface. If a product is felt 6 1 telephony to tele-commuting. to be very desirable or very practical, 7 8 users are wont to put up with rather 9 0 The third root of human factors is more mediocre to bad user interfaces, but recent, viz. the advent of the electronic when the product has received a wide digital computer. At first, the big main- acceptance, demands for better usability ITU-T E.161 layout of dial used frames were operated by specially trained soon arises in Norway except Oslo staff, but with the advent of the micro- computer (the PC, and not least the Mac) computers got into the hands of all kinds learning to organisational processes and of people. In the beginning, cryptic com- so on, requiring cross-sector qualifica- mands and inscrutable operating systems tions and excellence far beyond basic created great problems for ordinary psychology [13,14]. people using this new technology. This 6 7 gave rise to HCI (Human Computer As deregulation, liberalisation and privat- 5 8 Interaction) and the HCI research led to isation of the telecommunication sector 4 9 the graphical user interfaces (GUI) with has swept across Europe, the new-found 3 windows, menus, buttons, icons, pic- competition has led to the prices, which 2 tograms, and object manipulation interac- were previously determined by political 1 0 tion. This started in the late 70s, but rather than market factors, now coming gained momentum during the 80s and is down. As prices are further pressed down now a well established discipline. As by the increasing competition, operating Layout of dials used in Oslo telecommunications and IT are merging, companies have to look to other areas to so are also the fields of human factors in gain a competitive edge. And the most telecommunications and HCI until they current area here is usability, i.e. how now have become inseparable. user-friendly a service or product is.

We are long past the days when young Why human factors? computer buffs found a challenge in every new service or way of getting Human factors is based on thorough in- 3 2 around in what we now call the Internet, 4 1 depth knowledge about the physical and 0 trying out modems and new ways of 5 mental (i.e. cognitive) abilities of human getting around obstacles in the net. Users 6 users to address various areas in telecom- 7 today demand fully ‘transparent’ ser- munication. This may range from specifi- 8 9 vices, i.e. they do not want (and should cation of new services to improving not need) to concern themselves with existing services, from user testing to how a connection is established, whether writing style guides, from synthetic terrestrial or by satellite, whether by Layout used in Sweden speech and speech recognition to design analogue or digital transmission, whether of web-sites, from the psychology of line switched or packet switched,

Telektronikk 3/4.1997 5 whether over PSTN or ISDN [13,14]. They also demand one-stop-shopping, i.e. to deal with only one company to deliver all that is necessary for a service in one bundle rather than having to deal with a number of different vendors to obtain line access, modem, terminal, communication software, the right con- nectors and all the other bits and pieces necessary to get a service up and running. The user simply wants to dial a subscriber number to get in touch with the desired party, to type out an e-mail address, to send an e-mail or a web address, to reach a particular home page. Usability will, therefore, become the most important competitive issue after prices have come down. This trend has already been recognised by the larger and more foresighted operators, but other operators are quickly following suit to exploit the commercial advantages offered by improved usability. Figure 7 A not-so-funny situation for many users of modern electronic consumer If subscribers are to choose a particular equipment. It is disheartening to see that so few producers actually take the needs of service provider it is important that the the users into account when designing new consumer equipment. The domestic video- operator provides services and products recorder (VCR) has been a particularly troublesome product which do not require the subscriber to (Cartoon by Martin Holmes; from [15].) have to memorise long non-intuitive procedures, codes or number strings, or to have to read and understand fat operating manuals. Professor Peter Cochrane of BT Labs in the UK has proposed the ultimate user-friendly paradigm for future ICT (Information and Communi- cation Technology, a term coined by Commissioner Bangeman of the CEC): “Three clicks”, “One second” and “No manual”; which means that you should be able to access a desired file or service by not more than three mouse clicks (or three menu choices), that no action should take longer than one second before the user gets a response, and that there should be no need to read user manuals or instructions to be able to use a service or function. This may sound rather utopian, but it is a desirable goal to work towards.

Both the technologies and the necessary human factors knowledge are, by and large, already available, but to achieve many of these desirable goals it will be necessary for some cross sector co-operation; such as between telecommunications engineers, psychologists and programmers and others involved in designing new services and products.

Figure 8 ‘Over-featuring’ is a vice often indulged in by the producers of consumer A current hot topic is what is called equipment and it invariably leads to ‘under-use’: Most of all those clever functions ‘Design for all’, ‘Universal design’ or that are invariably crammed into modern communication products are seldom, if ever, ‘Designing for our future selves’, used. A bad case of ‘creeping featurism’(Cartoon by Martin Holmes; from [15].) implying that when new products and

6 Telektronikk 3/4.1997 services are designed these must be usable by as many people as possible. To achieve this it is paramount that human factors experts are brought into the design process as early as possible. The importance of paying attention to human factors issues at an early development stage has been stressed many times, but cannot be repeated too often. Neglect of this may lead to unusable products and services and loss of customers and revenue, requiring costly corrective measures to satisfy the expectations and needs of today’s users.

More people live longer, in better health and with more money to spend than at any time before in our history. The market is changing fast as a result of this changing demographic age distribution: As a result of improved nutrition, health care and economy, the population is ageing fast, meaning that a larger and larger proportion of the population will be made up of older people. This development is especially marked in the industrialised countries in Europe and North America, but also in the developing countries of the Third World we see the same trend. The average life expect- Figure 9 The interactive telephone stored voice services can sometimes be a ancy is not always a valid indicator of the nuisance to some users growth of the ageing population since it does not take account of the increasingly skewed age distribution. We are not only talking about the young elderly (i.e. people in their late 50s or 60s, nor about and running a usability lab that in its what over-enthusiastic philanthropists. the middle aged elderly (i.e. those in their simplest form may consist of video Today we know that substantial eco- 70s and early 80s, but we are talking recording equipment, a computer and nomic gains can be derived from invest- about the older elderly (i.e. those in their various software packages. The cost of ing in human factors during product late 80s and 90s) who will also be users this lab, which does not have to be very development. The figures cited may vary of modern ICT services. These user fancy or expensive, is spread over the somewhat, but conservative estimates groups cannot be neglected any longer various development projects. The vari- indicate gains of at least two times the without dire consequences for the service able costs lay mainly in applying the investment – not a bad return on your providers. Now when the power of the method(s) most appropriate for a specific money by any standard. Most trade jour- state monopolies have been broken the product during the early development nals and popular journals now regularly subscribers do not have to put up with phase, e.g. heuristics, ‘Wizard of Oz’, publish test reviews where user interface whatever service or product the mono- cognitive walk-through, user involve- and general usability of products are poly offered. The users will now have ment, or any combination of these assessed. Buying behaviour is increas- real choices, and may not want to spend methods. ingly being influenced and dictated by their money on services and equipment such product reviews – a product that they cannot use. Thus, it becomes more The cost of applying such methods can receives a negative test review stands lit- important than ever to create services and now be estimated quite reliably. The ben- tle chance in the marketplace. products adapted to the special abilities efits gained by increased sales or by and needs of the various elderly users [2]. more efficient use of a product with a well-designed user interface can also be Bell – the first human The cost of user estimated with some confidence. By sub- factors expert on tele- tracting the costs from the gains, the net phones! friendliness gain of developing and applying a good user-interface is derived [16,17,18]. When Alexander Graham Bell and Good user-friendliness does not come Thomas Watson in 1876 developed the free and the cost of creating a good user The application of human factors in first useful system for transmitting the interface should not be underestimated. interface design may have been regarded human voice over a wire, Bell did this Ignoring this issue may have dire conse- as a form of esoteric and costly charity, out of a deep dedication to improve com- quences. The fixed costs lay in equipping advocated by well-meaning, but some-

Telektronikk 3/4.1997 7 NORWEGIAN TELECOMMUNICATIONS MUSEUM

References 9 ETSI. Generic user control procedures for telecommunications ser- 1 ETSI. Access to telecommunications vices and terminals. ETSI Technical for people with special needs : Re- Report ETR 170, 1995. commendations for improving and adapting telecommunication termi- 10 ITU. Arrangement of letters, digits nals and services for people with and symbols on telephones and other impairments. ETSI Technical Report devices that can be used for gaining ETR 29, 1991. access to a telephone network. ITU-T Recommendation E.161 (03/93), 2 ETSI. The implications of human 1993. ageing on the design of telephone terminals. ETSI Technical Report 11 ISO. Information Technology : key- ETR 334, 1996. board layouts for text and office systems : Part 4 : Numeric Section First 3 Gagliardi, D. Report on the audible Edition. ISO 9995-4, 1994. tones in the telephone service in the EC countries. Report from contract 12 ETSI. Human Factors (HF) : charac- 48159 DGXIII, European Commis- teristics of telephone keypads and Figure 10 Alexander Graham Bell (1847 sion, Brussels, 1993. keyboards : Requirements for elderly – 1922), who together with Thomas and disabled people. ETSI Technical Watson invented the first usable tele- 4 Anderson, D M, Ferris, A, Mellors, Report ETR 345, 1997. phone in 1876, had also studied the W. Human factors in ETSI : use of potential users’ abilities to operate the tones in public telephone networks. 13 Norman, D A. The psychology of telephone system. Knowing well from his In: K. Nordby (ed.). Proceedings of everyday things. New York, Basic work with deaf people how difficult it can the 16th International Symposium on Books, 1988. (ISBN 0-465-06709-3.) be to use technical aids, he wanted the Human Factors in Telecommuni- new telephone to be easy to use, thus cations, Oslo, 12 – 16 May, 1997. 14 Norman, D A. The design of every- making him the first ‘human factors in (ISBN 82-994236-0-0.) day things. New York, Doubleday/ telecommunications specialist’ Currency, 1989. (ISBN: 0-385- 5 ETSI. European harmonization of 26774-6.) network generated tones. Part 1: A munication for deaf people (see Figure review and recommendations. 15 Collins, S C, Rabbitt, P. Age and 10). It is, therefore, a deeply tragic irony DTR/HF-01026-1, 1997. Design for new Technology. Age & that no other radical technological inno- Performance Research Centre, Uni- vation has discriminated against deaf 6 ETSI. European harmonization of versity of Manchester, 1990. people to such an extent as the telephone network generated tones. Part 2 : has. Bell’s studies in the physiology of Listing and analysis of European, 16 Mayhew, D J, Mantei, M. A basic speech and hearing in man provided the World and standardized tones. framework for cost-justifying usabil- basis for his invention, but he did not DTR/HF-01026-2, 1997. ity engineering. In: R G Bias & D J only study technology and physiology, Mayhew (eds.). Cost-Justifying he also studied the potential users’ abili- 7 ETSI. Minimum man-machine inter- Usability. Boston, Academic Press, ties to operate the new system. This actu- face (MMI) to public network based 1994. (ISBN 0-12-095810-4.) ally makes Bell the first human factors supplementary services. ETSI ‘expert’ in telephony. However, as the Telecommunications Standard ETS 17 Harrison, M C, Henneman, R L, technological development gained 300 738, 1997. Blatt, L A. Design of a Human Fac- momentum, the users’ requirements soon tors cost-justifying tool. In: R G Bias had to take a back seat. It is important to 8 ETSI. Guide for usability evaluations & D J Mayhew (eds.). Cost-Justify- get the user back in the front seat again, of telecommunications systems and ing Usability. Boston, Academic which would also be very much in line services. ETSI Technical Report Press, 1994. (ISBN 0-12-095810-4.) with Bell’s original intentions. ETR 095, 1993. 18 Bias, R G, Mayhew, D J (eds.). Cost- Justifying Usability. Boston, Aca- demic Press, 1994. (ISBN 0-12- Knut Nordby is Senior Research Psychologist at 095810-4.) Telenor R&D, Kjeller, where he has worked since 1985. He is Chairman of the ETSI Technical Committee Human Factors (TC HF), of which he was a founding member. He is member of the Permanent Steering Committee of Human Factors in Telecommunications (HFT) and Adjunct Professor teaching Man-Machine Interaction at UNIK, Center for Technology at Kjeller, University of Oslo. e-mail: [email protected]

8 Telektronikk 3/4.1997 Consumers, not Clients – Empowering users of new technology FRODE VOLDEN

Although new technology has given and the results indicate a significant bilities of individuals with disabilities. many individuals with disabilities new impact of AT on many aspects of the (20 U.S.C. Chapter 33, Section 1401 opportunities, many of the possible respondents’ lives, including: the (25)) [3]. users of such technology do not benefit majority of infants with disabilities bene- from these new opportunities. Is this fited by having fewer health problems This definition is broad and includes a due to poor availability of expertise because of AT; nearly 75 % of school range of devices from low technology to and technology, or may the hindrance age children were able to remain in a reg- high technology items as well as soft- be certain qualities of the service de- ular classroom, and 45 % were able to ware. livery system? Instead of focusing on reduce their use of school-related ser- availability of expertise, there are a vices; 65 % of working-age persons were Under IDEA the legal definition of number of reasons why we should able to reduce dependence on family assistive technology services is: rather focus on the users’ knowledge members, 58 % were able to reduce ... any service that directly assists an and skills. Decisions taken by the per- dependence on paid assistance, and 37 % individual with a disability in the son himself rather than by others, were able to increase earnings. Among selection, acquisition, or use of an increase the motivation to succeed in elderly persons, 80 % were able to assistive technology device. (20 U.S.C. using new technology. A shift of focus reduce dependence on others, 50 % were Chapter 33, Section 1401 (26)) [3]. from expert-driven decisions to user- able to reduce dependence on paid per- driven decisions also has implications sons, and 50 % were able to avoid enter- Assistive technology can be simple or for other groups of new technology ing a nursing home. complex. It includes Velcro, adapted users, such as elderly people and clothing and toys, computers, telephony groups in the society that traditionally aids, seating systems powered mobility, have been reluctant to take advantage What is augmentative communication devices, of new technology. Using the term con- assistive technology? special switches, assisted listening de- sumer rather than the term client may vices, visual aids, memory prosthetics, be helpful. There are a number of definitions of the and thousands of other commercially terms assistive technology and assistive available or adapted items. These tech- technology services. A simple definition New technology and services, aimed at a nology solutions improve an individual’s can be: technology for more independent, growing market of disabled persons, give ability to learn, compete, work and inter- productive and enjoyable living. A more many individuals new opportunities, and act with family and friends. People use comprehensive definition can be found in access to information previously un- assistive technology to achieve greater the American Individuals with Disabili- reachable. Estimates indicate that some independence and to enhance the quality ties Education Act (IDEA). 10 million people in Europe are potential of their lives. users of assistive technology due to dif- IDEA defines an assistive technology ferent impairments. This includes per- device as: sons with dexterity impairments, visual Telecommunication and impairments, hearing impairments, and ... any item, piece of equipment or assistive technology communication impaired. About 70 % of product system, whether acquired this population are persons aged 60 and commercially off the shelf, modified, or Telecommunication equipment for the over, see Table 1. customised, that is used to increase, elderly or disabled can be simple addi- maintain, or improve functional capa- tions to existing equipment like enlarged Telecommunication equipment and services as well as other technology have increased many disabled persons’ opportunities for independent living. The number of potential users is huge, only a frac- tion of these are actual users of assistive technology today. Table 1 Estimated development of the disabled population in 11 EU Member states (thousands). Source: [1]. The impact & benefit of assistive technology 1993, 2020, 1993, 2020, Total Total 60 + 60 + A number of studies have shown that many individuals greatly improve several factors associated with quality of life and Dexterity impairments 2,307 2,829 1,645 2,141 independence by using assistive technology. As an example of this, a study [2] Visual impairments 2,189 2,727 1,732 2,254 surveyed 136 individuals with disabilities to evaluate the costs and benefits asso- Hearing impairments 3,419 4,223 2,528 2,290 ciated with the use of different kinds of technology-related assistance. The main Communication impaired 1,876 2,308 1,212 1,578 focus in this report was independence. The individuals were in four age groups Total 9,791 12,087 7,117 9,263

Telektronikk 3/4.1997 9 Expert – client “Big Button” A focus in (re)habilitation research has Big Button is a telephone key- been on finding a “perfect match” be- pad with extra large recessed tween user characteristics and assistive keys, and is used in addition to technology characteristics. This is of an ordinary telephone. It is course a necessary part of the service delivery, but probably not sufficient. designed to assist persons who have difficulties hitting the right Traditionally, the focus in the process of key due to spasms or trem- service delivery for disabled persons in bling, or to poor eye sight. need of assistive technology, has been very expert centred. The user has been It can also be programmed so met with “paternalistic” attitudes, and that it is enough to press one has been advised what technology he key. In this mode the numbers should use, and what to do. Patricia Thornton [5] states that although recent on the keypad may be replaced by symbols or pictures, making it possible for per- developments in communication technol- sons with different cognitive disabilities using telephone themselves. ogy can reduce dependence on others and facilitate independent living for the disabled, the dominant model of provision emphasises vulnerability, assumes a need for protection and imposes rules about appropriate use of the devices. She argues that this rather leads to disem- powerment than empowerment of the user. or adapted keypads for standard tele- hold appliances can be remotely con- phony, complex solutions for telework- trolled, and alarms can be set to alert the Too often the outcome is that people with ing and “smart-house”. Telecommunica- user or others when certain conditions disabilities have little choice about the tion services like text-telephony and are satisfied. services, programs, supports, or accom- speech-interfaces rather then keypad, are modations they may receive. Decisions also examples of telecom based equip- have already been made for them. The ment and services for disabled users. Why don’t all benefit? empowerment of choice is a key to achieving independence. The Internet has given many groups of It is pretty clear that assistive technology disabled people access to information can improve many individuals’ quality of Society has changed, and the traditional they previously had great difficulties life. What we often see is that persons we expert centred service delivery system accessing. This includes information thought would benefit from using tech- has of course changed with it. It is con- databases, newspapers, discussion nology, do not take advantage of these sidered a democratic right to be informed groups, etc. If the user can control and possibilities. Problems with availability about decisions concerning yourself. But, understand a computer using necessary and funding as well as available expertise this does not necessarily mean that the input and output devices, he can in most are of course major factors here. But, user is the one in control of the process. cases access all the information non-dis- funding and availability of technology is abled persons can access. hardly the problem in our part of the Disabled persons do not have more in world. common than other people do. They are The information-based society, supported individuals with their own personality, by information technology and an ad- Provision of assistive technology in Nor- interests, and priorities. Service delivery vanced communications infrastructure, way as well as the other Nordic countries systems that do not involve the user and will probably lead to flexible organisa- is a key part of our social security system others close to the user will have diffi- tions and new ways of working. An ex- and fully based on governmental fund- culties in fully adjusting to the user’s ample of this is the rise in the promi- ing. Basically, if you document a need needs and priorities. nence of teleworking. The disabled user for assistive technology, you get it. who controls a personal computer, will benefit from this development. Easy access to the technology does not The goal: Independence necessarily mean that more people take The concept “Smart-House” means an advantage of the technological possibili- through empowerment adaptation of a domestic environment ties. Several studies have shown a poor What we want to achieve when imple- thanks to computers and telecommunica- relationship between the number of menting assistive technology is to enable tions. The devices in the automated assistive devices provided and the the user to control his or her own life. “Smart-House” work together, and many number in actual use [4]. There is reason Independence is the keyword. The popu- functions in the house is remotely con- to believe that there are qualities with the lar term empowerment may be useful in trolled. For instance, doors and windows service delivery process itself which ex- this connection. Empowerment is about can open automatically, electrical house- plain this lack of relationship.

10 Telektronikk 3/4.1997 putting the individual in control, giving him or her the tools to control their own Using life. eye tracking In many countries, most noticeable in the to control USA, civil rights movements have focused on disabled persons’ right to inde- appliances pendent living and full integration in society. Taylor, Biklen, and Knoll [6] have tried to describe what independence and community integration for people with disabilities mean. They state that all people belong in community, regardless of their type or degree of disability, and community integration involves participation, community living skills, and advocacy. The individual will be asking questions along with professionals about what it means to be a full participant in the community. This implies that people with disabilities should be put in a position where they run their own programs.

Motivation

A key issue in the use of new technology is motivation, and an effective way of increasing motivation to take advantage of new technology is created through actual participation from the user in the decision process.

A good match between the user’s cognitive and physical characteristics, and technology is necessary, but not sufficient. Starting to use assistive technology is often difficult. In an introduction phase, the assistive devices may decrease The Eyegaze communication system rather than increase the user’s function. The Eyegaze system makes it possible for people with physical disabilities to operate a com- Learning to take advantage of the devices puter with their eyes. By looking at control keys displayed on a computer monitor, a person can often requires lots of effort from the user synthesize speech, control his environment (lights, appliances, etc.), type, operate a telephone, and from others near the user. If the run computer software, and access the Internet. selection of technical solutions was just an advice from an expert, it is easy to Who uses the Eyegaze system? disregard the expert’s competence and Eyegaze systems are used around the United States, Canada, South and Central America, and stop using the assistive device. If the user Europe. Their users are adults and children with cerebral palsy, spinal cord injuries, head in- himself selects the device, he probably is juries, ALS, multiple sclerosis, muscular dystrophy, brainstem strokes, and Werdnig-Hoffman willing to invest much more effort into syndrome. Eyegaze systems are being used in homes, offices, schools, hospitals, and long- learning to use the device and adopt term care facilities. usage of the device in everyday life. How does the Eyegaze system work? As a user sits in front of the Eyegaze monitor, a specialized video camera mounted below the Consumers, not clients computer monitor observes one of the user’s eyes. Sophisticated image-processing software in the computer continually analyzes the video image of the eye and determines where on the What we would like to achieve is putting computer screen the user is looking. Nothing is attached to the user’s head or body. the user in charge of his own service delivery process. One step on this road The Eyegaze system, developed and sold by LC Technologies, is an example of a “High-end” could be changing experts attitudes product that integrates a number of functions that may be useful for a disabled user, for in- towards our clients. Using the term con- dependent living. This is a complex system that requires effort both to learn and to maintain. sumer rather than client may be one way More information on this system can be found at their Internet site: of achieving this. http://www.lctinc.com

Telektronikk 3/4.1997 11 Nosek [7], in providing support for the word “consumer”, says that rehabilitation Individual has a paternalistic past and goes on to characteristics Posibilities suggest that rehabilitation is in the midst of a revolution toward a more humane and creative discipline that focuses on both independence and productivity. She Outcome suggests that use of the word consumer symbolises that progress and maintains that use of the word will destroy status quo. Consumer empowerment may be the primary focus of the future. Motivation Older members of society facing adjust- ments to ageing require services extending beyond traditional views of rehabili- Figure 1 Elements influencing outcome of a service delivery process tation. Merging the needs of the ageing of assistive technology population with those of disabled persons as advocated by Zola [8] adds strength to consumers’ roles. Universal policies recognising that the entire population is at risk for chronic illness and disability will prevent perpetuation of segregated, unequal parts of society. Recognition that Producers/ everyone with a disability will age, and Companies everyone who ages will acquire one or more disabilities is yet another challenge which broadens the view of those persons providing and receiving services. Client Experts Decisions Betty Friedan [9] captures the evolving spirit of the ageing in her recent book entitled ‘The Fountain of Age’. The older age group intends to have a voice regarding public policies and speak loudly on . Helpers issues that concern them. Life beyond fifty no longer means ill health and isolation. Older individuals receiving future rehabilitation services will not be, nor do a) they view themselves as, a minority or disadvantaged group. The present day strength of such groups as the American Association of Retired People (AARP) symbolises the power of this ever growing population segment. Experts Access to information

Consumerism requires information; thus, increased access to information accele- rates the trend toward rehabilitation con- Helpers Consumer Decisions sumerism [10]. The communication and technological revolutions have made this a simpler task. Persons with disabilities can access information independently Producers/ using computers and telecommunications Companies providing access to otherwise inacces- sible information [11].

As technological advances enhance b) opportunities for disabled people, rehabilitation counselling needs to keep pace Figure 2 The user should be in control of the service delivery process, in effectively matching consumers to rather than leaving the decisions to an expert

12 Telektronikk 3/4.1997 technological aids. These advances have 10 Inlander, C B. Trends in medical brought to the forefront the need for pro- consumerism. In: A M Rees (ed). fessionals to participate more extensively Managing consumer health informa- in rehabilitation assessment and training tion services. Phoenix, AZ, Oryx [12]. Press, 3–14, 1991.

Reed et al. [13] are advocating what they 11 Fullmer, S, Majumder, R K. In- call a consumer driven model for service creased access and use of disability delivery, and describe experiences from related information for consumers. projects that have tried to focus their ser- Journal of Rehabilitation, Jul/Aug/ vice delivery this way. What they call the Sep, 17–2, 1991. “local resource team model”, is a model they see as well suited for a consumer 12 Hale, P N Jr, Shipp, M. Personal driven approach. transportation : a challenge to the rehabilitation community. Journal of Rehabilitation, Apr/May/Jun, 8–9, References 1989. 1 Carruthers, S, Humphreys, A, 13 Reed, B J, Fried, J H, Rhoades, B J. Sandhu, J. The market for R.T. in Empowerment and Assistive technol- Europe : a demographic study of ogy : the local resource team model. need. In: B Ballabio et al. (eds). Journal of Rehabilitation, 61, (2), Rehabilitation Technology. Amster- 30–35, 1995. dam, IOS Press, 1993.

2 National Council on Disability. Study on the financing assistive technology devices and services for individuals with disabilities : a report to the President and Congress, 1993.

3 The Individuals with Disabilities Education Act (IDEA), 1997.

4 Wormnæs, S. En større verden : datateknologi for personer med bevegelses- og talevansker. Oslo, Universitetsforlaget, 1992.

5 Thornton, P. Communication technology : empowerment or disempow- erment. Disability, Handicap and So- ciety, 8, 4 Dec, 339–349, 1993.

6 Taylor, S J, Biklen, D, Knoll, J (eds). Community integration for people with severe disabilities. NY, Teachers College Press, 1987.

7 Nosek, M A. A response to Kenneth R Thomas’ commentary : some observations on the use of the word “consumer”. Journal of Rehabilita- tion, Apr/May/Jun, 9–10, 1993. Frode Volden is Research Psychologist at Telenor 8 Zola, I K. Ageing and disability : Research & Development, Kjeller toward a unified agenda. Journal of e-mail address: Rehabilitation, Oct/Nov/Dec, 6–8, [email protected] 1989.

9 Friedan, B. The fountain of age. New York, Simon & Schuster, 1993.

Telektronikk 3/4.1997 13 Speech quality, usability and service quality NORMAN GLEISS

. 1 The concept of speech C = B log2 (1 + S / N) bit/s connection with a bandwidth of B = 300 – 2800 Hz and a signal-to-noise ratio of transmission quality There is experimental evidence that S/N = 16 dB. These values would, how- speech quality Q is indeed a function of ever, be outside the range recommended The perceived sound quality of speech the speech and noise levels and of the by ITU-T for international telephone transmitted and reproduced over a tech- bandwidth, i.e. Q = f(S, N, B). The pos- transmission, and the quality of such a nical system, in particular the telephone sible relation to the transmission of in- connection would be judged as very bad. system, may be described by a number of formation has been investigated by Actually, a capacity of 50–70 kbit/s is attributes, which ultimately affect the Richards and Swaffield [1]. Suppose that needed for good speech transmission. On usability of telephony services. These the information H bit/s is fed into the the other hand it is now possible to attributes may be derived from technical, i sending end of a speech transmission link obtain satisfactory speech transmission physiological or psychological consider- having a capacity of C bit/s, and that the quality at 16 kbit/s or even less by ad- ations and be based on empirical data amount H reaches the receiving end vanced digital coding techniques. from these domains. o each second. With insufficient channel capacity, i.e. C < Hi, we get Ho / Hi <1, A different kind of limitation is the rate 1.1 Information theory but when C > Hi we have Ho = Hi. at which the brain can handle the in- A high-quality circuit could have a formation coming from the ears. This is Technically, some kind of correlation certain excess capacity C – Hi. With very much smaller than the hearing reso- may be expected between speech quality these assumptions, the ratio of Ho / Hi as lution: only of the order of 10 bit/s. It and the ratio of the amount of inform- a function of channel capacity would should therefore be possible to transmit ation transmitted over a specific system follow the curves shown in Figure 1. at least the semantic information con- to the total information content of the Here the linear theoretical function pre- tained in a speech signal over a channel speech spectrum. If this ratio is <1 it is sumes a perfect coding and decoding of with a considerably lower bit rate, if a reasonable to believe that speech quality the information stream. The practical suitable coding procedure could be is degraded; at least that should be true performance will follow the dashed curve found. In fact, a telex system having a for analogue signal transmission, dis- because of the less than ideal electro- capacity of 50 bit/s is capable of trans- regarding the redundancy of speech acoustical coding in the telephone set. mitting the semantic content of a mes- information and the possibility of intelli- sage in the form of text at about the same gent coding techniques which might However, the channel capacity required speed as the words are spoken. (Speech reduce the redundancy in a way that is for speech transmission is actually deter- in a language composed of 25 = 32 not noticeable to the listener. mined by the rate at which the hearing phonemes (5 bits) and spoken at a speed sense can resolve the acoustic inform- of 10 phonemes per second implies a According to Shannon’s formula, the ation received rather than by the inform- transmission rate of 50 bit/s.) maximum amount of information, C, ation content of the input signal. The which can be transmitted over a channel, maximum receive rate is estimated to be 1.2 The hearing sense is proportional to the bandwidth B, the 15 kbit/s. A value of C of this magnitude speech level S, and, inversely, to the corresponds to an analogue telephone The hearing sense has two characteristics noise level N: which are of particular importance for the perception of speech over a telephone system. One is the sensitivity in relation to frequency, which can be described by “isophones” pre- Ho/Hi senting the sound pressure level required to give a spe- "Poor" "Fair" "Good""Excellent" "Perfect" cific loudness sensation as a function of frequency. The 1 threshold of hearing is the lowest curve of the isophone family, corresponding to zero loudness. The threshold moves upwards in level when audible noise is added and becomes then a masked threshold.

C - Hi The other important characteristic is the loudness growth function, which presents the C perceived loudness of a sound 0 as a function of sound pres- Hi bit/s sure level. This function has a shape that to some degree Figure 1 Performance of a speech transmission link as a function of channel capacity. Input information depends on the spectrum of rate = Hi; output information rate = Ho. Full line: theoretical function; dashed line: practical performance

14 Telektronikk 3/4.1997 dB the sound. The loudness growth function for speech is a special case which will be discussed later in this paper.

A With consideration to these characteris- Area of perceivable speech tics of the hearing sense, the useful part of the information in transmitted speech may be expected to be that part of the speech spectrum which is left over after

being affected by the loss, the noise and Level the band limitation introduced by the S transmission system. Figure 2 shows this o part as the area of information content in a level/frequency diagram, where the N S lower contour is defined by the hearing threshold of the listener and includes the H masking effect of supra-threshold noise. B

From this kind of presentation it is pos- Hz sible to calculate the intelligibility of the received speech in dependence of the Frequency quantities B, S and N. The calculation of Figure 2 Principles of the calculation of intelligibility from the received speech Articulation Index is the best known spectrum (S), hearing threshold (H), and noise level (N). The area of perceivable method for this purpose. Here the useful speech has the received speech spectrum (the sum of the input speech spectrum, spectrum is divided into a number of S and the frequency-dependent attenuation in the transmission channel, A) as frequency bands that each contribute to 0 the upper boundary and the masked hearing threshold (the sum of the threshold articulation or intelligibility by a certain and the noise level) as the lower boundary weighting coefficient that can be determined empirically. This method was originally proposed by Collard (1930) and has later been developed by, among others, French and Steinberg (1947). 1.3 Perceptual dimensions sions, and subjects were asked to judge the overall quality on an opinion scale. This acoustical description, involving the A large number of experiments have By different methods of statistical analy- physiological characteristics of speech been carried out over the years on ana- sis, in particular factor analysis, it has production and hearing, is also used in logue telephone transmission systems in been possible to identify the most impor- recent models for calculating not only order to investigate the relation between tant factors which make up the general intelligibility but also the overall per- physical system characteristics, basically concept of speech quality. ceived quality of speech impaired by a B, S and N, and perceived quality. In gen- transmission system. These models there- eral, these experiments have been con- Not surprisingly, the dominant factor is fore include empirical data from subjec- versation or listening tests where the sys- Loudness. In fact, most of the engineer- tive tests where the quality is rated on tem parameters were varied in all dimen- ing effort during the first 50 years of the scales representing different perceptual dimensions. There are furthermore specific models which are used to calculate only the loudness of speech. The results Table 1 Relation between the perceived quality of speech and the physical system may be presented as “Loudness Ratings”. characteristics In this case the non-linear function relating the growth of loudness with Physical factors Perceptual factors speech level is incorporated in the model. Overall quality Loudness Intelligibility Naturalness An overview of calculation models by which perceived dimensions of speech Speech level +++ +++ +++ + can be calculated from electro-acoustic data, basically measurements of fre- Noise level +++ + +++ + quency response (bandwidth) and speech Bandwidth +++ ++ ++ +++ and noise levels, is given in ITU-T Supplement 3 to the P-series Recom- Distortion ++ + ++ mendations [11]. Further work in the development of calculation models for Sidetone + +++ (Talker sidetone) (Listener sidetone) speech quality, including also the effects of digital coding, is presently going on in Delay ++ + ITU-T/Study Group 12. One recent result Echo ++ + + is Recommendation G.113 [9].

Telektronikk 3/4.1997 15 frequency characteristics or bandwidth. Overall Non-linear distortion, sidetone, delay and quality echo effects can be considered as secondary factors which contribute to a minor degree to the overall quality. The relation between the physical and perceptual I II III factors may be represented as in Table 1. Loudness Intelligibility Naturalness The strength of the individual relation between factors is approximately as indicated by the number of crosses. These IV relations are discussed in more detail in

Perceptual factors "Echo the next section. disturbance" The structural relationship between the perceptual and physical factors is illustrated in Figure 3. Apart from the basic factors S, N and B, which all have a SNBa b c d e f g h i j k• • • specific relation to the three most im- Physical factors portant perceptual dimensions, there may be a large number of other factors which Figure 3 Structural model of perceived speech quality and its components either affect the same dimensions or only one specific additional dimension. For example, talker echo means that a speaker hears his own voice again after a history of telephony has been devoted to factor which may be called “Natural- certain delay and with a certain attenu- produce a sufficiently high listening level ness”. This factor has to do with the non- ation (the “echo loss”). The perceived at the receiving end of the telephone con- semantic content of speech and involves disturbance of an echo is a complex nections. The second factor in impor- personal characteristics such as the iden- function of echo loss and delay and may tance is Intelligibility. Again this would tity and the state of mind of the speaker. form its own quality component, desig- be expected, because the principal ob- nated by IV in the figure, because it may jective has been to transmit intelligible The basic physical factors are, as ex- not significantly affect either loudness, speech. Only as number three comes a pected, speech level, noise level and intelligibility or naturalness.

Quality Loudness

Excellent 5 Much too loud 5

Good 4 Too loud 4

Fair 3 Preferred 3

Poor 2 Too quiet 2

Bad 1 Much to quiet 1

0 10 20 40 dB Loss 0 10 20 40 dB Loss

Figure 4 a) Speech quality; b) Loudness, as functions of transmission loss (decreasing speech level); general principle

16 Telektronikk 3/4.1997 2 Quality dimensions 5 5 2.1 Loudness of speech

Loudness is generally the dominating Q factor for the perception of speech 4 4 quality. A typical curve relating perceived quality to loss in a telephone transmission system is shown in Figure L (a) 4a (increasing loss means decreasing sound level at the listener’s ear). The 3 3 sound quality has a maximum at a spe- Quality cific sound level (corresponding to a Loudness certain loss, or “Loudness Rating”, in the system). The perceived quality falls off at (b) both sides of the maximum, although 2 2 steeper in the direction of increasing loss, coming down to zero when the speech is no longer audible. Optimum level Preferred level At the two sides of the maximum, points 1 1 of equal quality may be found; obviously one is too loud and one is too weak, and the optimum is somewhere in between. 0 -20 -30 -40 -50 dBm For further analysis, the Quality function may be compared with a similar function Speech level presenting loudness against loss (Figure Figure 5 a) Speech quality; b) Loudness, as functions of 4b). It might be expected that the opti- decreasing speech voltage level, as obtained in a listen- mum speech level (at “optimum” loss) ing test with 20 subjects over a modern telephone set coincides with the most comfortable or preferred listening level, and in fact in earlier literature the diagrams are often drawn in that way (e.g. [3]). However, the results of tests where subjects are (with a certain noise level). The same region compared to the ends appears to asked to rate quality and loudness inde- level of 65–70 dB is considered as a be typical for speech and can be ex- pendently in a modern telephone set, may comfortable level when listening to a plained as a psychological “constancy look as in Figure 5. Similar results, con- loudspeaker telephone. One reason for effect”: a change of sound pressure in the tributed to Study Group 12, have been this difference of about 15 dB is that the range where conversational speech is obtained in different laboratories. There- preferred listening level increases with normally presented to the ear is to some fore, the curves in Figure 4 demonstrate a room noise, but the greater part can be extent perceived as a change in distance typical effect, implying that the optimum derived from the difference between to the talker rather than in loudness. The speech level is significantly higher than monaural and binaural listening, see Fig- perceptual image is similar to that of a the preferred listening level. Differences ure 6. In the figure, the estimated loud- talker changing the spatial distance to a in the range of 3–8 dB have been found, ness for continuous speech is presented listener in the same room while main- depending, among other things, on the on a logarithmic scale as a function of taining his speaking level; in a real con- type of telephone. This effect may be the sound pressure level at the ear for one versation situation the listener would interpreted to mean that maximum and two ear listening, as measured by correctly interpret the change of sound speech quality is perceived to occur at a Gleiss and Goldstein [4]. It can be seen level at his ear as a change in distance higher level than the most comfortable that in order to produce the same loud- and not in loudness, whereas in a tele- listening level. Similar results have been ness for monaural listening as for binau- phone conversation the same interpreta- reported for users of hearing aids. ral listening, the level has to be raised by tion becomes an illusion. (For earlier about 12 dB. Therefore, it is very likely generations of telephone users it was per- The global telephone network is designed that the above mentioned difference of haps reality, because, in the analogue so that as many of the users as possible 15 dB can mainly be attributed to the network, loss generally increased with get a listening level that is close to the monaural/binaural difference. On the the geographical distance between the optimum level. With national standard other hand, going from one to two ear lis- partners in a conversation and so con- telephone sets complying with ITU-T tening with the same level at the ear tributed to the sensation of physical dis- Recommendations the sound pressure appears to increase the loudness only by tance.) This effect enhances the impor- level will then be in the range of 1.7 : 1, while earlier estimates have been tance of being in the vicinity of the opti- 80–85 dB. This value is considerably 2:1. mum listening level in order to give an higher than the average listening level, impression of social presence to the which is about 65–70 dB, during a con- The shape of the curves in Figure 6 partners in a telephone conversation. versation between two persons at a con- resembles the loudness curve in Figure venient distance in an ordinary room 4b. The reduced slope in the middle

Telektronikk 3/4.1997 17 103

a) Binaural

102 Magnitude estimation (log Loudness) b) Monaural

101 40 50 60 70 80 90 100 110 120 dB

Sound Preassure Level (SPL) Figure 6 Loudness as function of speech level at the ear, for a) two ear, and b) one ear listening (curve fitted to data from Gleiss and Goldstein)

Quality 4.0

3.5

3.0 Increasing noise level

2.5

2.0 Mean opinion score 1.5 - 80 dBmp - 70 dBmp 1.0 - 60 dBmp - 50 dBmp - 40 dBmp 0.5

0 0 4 8 12 16 20 24 2832 36 40 dB Overall loudness rating Loss Figure 7 Perceived quality, as mean opinion scores on an overall quality scale, as a function of transmission loss (decreasing speech level) with noise level as a parameter

18 Telektronikk 3/4.1997 100

60 Key words in simple sentences

Small closed sets of two words 50 Small closed sets of six words

Test vocabulary of

Intelligibility (%) 40 256 monosyllabic words

30 Monosyllabic words (1000 different words)

Mixture of monosyllabic 20 logatoms and words (600 - 1000 different syllables) 10

0 -12 -9 -6 -3 0315186 9 12 Speech to noise ratio (dB)

Figure 8 Intelligibility as a function of speech-to-noise ratio (S/N) for various types of speech material (From ISO)

Another effect to consider in the plan- in this diagram). At least over a limited well as of the transmission system. The ning of telephone systems is the in- range (on one side of the optimum and expectations of the listener and the infor- fluence of noise. Line noise as well as for higher noise levels), these isoprefer- mation content of the speech material room noise can reduce the signal-to-noise ence contours will correspond to constant presented in relation to the listener’s own ratio for speech at the listener’s ear. Con- S/N values. vocabulary have a strong influence on sequently, intelligibility may also be the measured intelligibility. However, reduced, and therefore the optimum 2.2 Intelligibility with this in mind it is possible to at least listening level in the presence of noise rank order different systems in quality may be shifted towards higher speech with a given listener group and a specific Intelligibility, or articulation1, is gener- levels. This effect is illustrated in Figure type of speech material. On the other ally measured by counting the proportion 7, showing a family of curves with hand, in order to come to conclusions on of correctly understood sounds, syllables, different noise levels as parameters. The an absolute basis, the average perfor- words, or sentences in a listening test uppermost curve without noise is in mance of a large number of listeners has with a group of subjects. In principle, the principle the same as in Figure 4a. It is to be assessed. results for a system without bandwidth clearly seen that for higher noise levels, limitation or other distortion will come the optimum points are shifted to the left, Formal articulation tests are rarely used out as in Figure 8. The curves describing i.e. to higher listening levels (lower loss nowadays, not only because they are time intelligibility as a function of S or S/N values). As a consequence, “isoprefer- consuming but even more because trans- have a sigmoid shape going from 0 % ence” contours can be found which con- mission systems in general are required and gradually approaching 100 %. They tain points on different curves having the to be well outside the saturation level of are centred around the 50 % level, which same perceived quality (horizontal lines 100 % of the curves in Figure 8, in fact may be called the threshold of intelligi- outside the range covered in the diagram. bility. These general results mean that In practice, telephone systems should intelligibility primarily depends on the 1 In technical language, articulation have an intelligibility of at least 90 % for signal-to-noise ratio but also on the usually refers to identifying sounds syllables and almost 100 % for sentences. speech material itself. Intelligibility tests and nonsense syllables, while intelligi- It is therefore not possible to differentiate are in fact performance tests, which mea- bility refers to understanding words between actual systems by articulation sure the performance of the listener as and sentences. tests. When new systems or components,

Telektronikk 3/4.1997 19 Suitable rating scales for this purpose 2.3 Naturalness may have a zero point labelled “No meaning understood with any feasible It has already been mentioned that the effort”, while the upper end of the scales concept of naturalness covers speaker may be labelled “Complete intelligi- recognition and the timbre of the voice

Naturalness bility” or “Complete relaxation possible; but also other more subtle non-semantic no effort required”. components such as the perception of the speaker’s emotional status as reflected in Intelligibility is also dependent on the the character of the voice. Recent experi- frequency response or bandwidth of a ments have shown that facial expressions f 1 transmission system. The conventional like smiling or frowning not only affect 180 telephone band is limited to 300 – the acoustic speech spectrum in a mea- Upper band limit f 2 3400 Hz. Historically, the band limits surable way but also that listeners can 225 3500 were chosen as the frequency limits for detect and identify these expressions 280 low-pass and high-pass filters which from the voice character without seeing 2800 gave the same reduction in intelligibility. an image of the speaker. On the other Hz Lower band limit Extending the band limit upwards in hand, seeing the face of the talker, as in frequency does not dramatically improve videotelephony, can significantly intelligibility. An extension up to improve intelligibility, at least under Figure 9 Perceived naturalness of 7000 Hz, which is the standardised upper adverse acoustical conditions, and also speech transmitted over frequency bands limit for ISDN broadband telephony, contributes positively to the perceived with different upper (f ) and lower (f ) 1 2 certainly increases the intelligibility for speech quality. cut-off frequencies. Results from paired some consonants, especially fricatives comparison listening tests; male and like f and s, but at the same time the The frequency response of a speech female voices relative improvement in overall sound transmission system affects intelligibility quality is much more noticeable. This and naturalness in different ways. Natu- improvement in quality is even more ralness is in most applications improved pronounced when the frequency band is by a full bass reproduction and also a extended downwards. It appears that certain degree of treble boost, particu- changes in bandwidth, at least outside the larly if the upper cut-off frequency of the e.g. speech coders, are introduced it is conventional telephone band, affect the transmission band is not sufficiently generally sufficient to rate “intelligi- factor of naturalness much more than high. Intelligibility, on the other hand, bility” or “listening effort” directly. This intelligibility. Naturalness should there- may be improved by a bass reduction and can be done with surprising accuracy and fore be treated as a separate issue, with a further emphasis on higher frequencies. with much less effort than articulation equal importance for speech quality as In general, naturalness seems to be more tests. intelligibility. dependent on the presence of the fundamental frequency (F0) and the first for- mant (F1) in the reproduced speech spectrum, while intelligibility is more dependent on the higher formants in the Overall voice spectrum. However, these effects speech depend to a high degree on the speakers’ quality 100 - 7000 Hz individual voice spectra. In particular, there are significant differences between male and female voices as regards the 200 - 3400 Hz magnitude of the effect of bandwidth restriction on speech quality. 300 - 3400 Hz A detailed analysis of the effect of changes in bandwidth and frequency response based on extensive tests at the

Intelligibility former Swedish Telecom Administration was made by Gleiss [2]. It was found that shifting the band limit by 100 Hz at the lower end affects naturalness much more than a similar shift at the upper end. Therefore, for example the frequency band 200–3050 Hz is perceived as clearly superior to the equally wide band 300–3150 Hz. If instead the limits of the frequency band are moved in steps of one Naturalness third-octave, results are obtained as Figure 10 Naturalness and intelligibility ratings for three different speech trans- shown in Figure 9 for six different bands. mission bands and their projection on a hypothetical quality scale It appears that the impairment caused by

20 Telektronikk 3/4.1997 decreasing the upper cut-off frequency from 3580 to 2800 Hz can almost be compensated for by decreasing the lower cut-off frequency from 280 to 380 Hz, A → A i.e. a decrease of 100 Hz at the lower end compensates a decrease of 750 Hz at the upper end with respect to the naturalness A → R of the transmitted speech. A conclusion is that the conventional telephony band of 300–3400 Hz is not the optimum choice and a displacement of the band 100 Hz downwards would improve the R → A speech quality significantly.

Further examples of the effects of spe- Intelligibility cific modifications in the frequency R → R response of a telephone system are given in Figures 10 and 11. Figure 10 shows how three different transmission bands Naturalness are positioned in a diagram where the co- ordinates are Intelligibility and Natural- Send → Receive ness. It is seen that an extension of the Flat response band downwards from 300 to 200 Hz A = Anechoic room increases naturalness but has no notice- R = Reverberant room Preemphasis 6 dB/oct. able effect on intelligibility. Increasing the upper band limit from 3400 to Figure 11 Naturalness and intelligibility ratings for a 7000 Hz (approximately one octave) loudspeaker telephone system in different room acoustics. makes a larger contribution to intelligi- The send and/or receive ends are in anechoic or reverber- bility but also to naturalness. ant room conditions. The microphone frequency response curve is either flat or has a 3 dB/octave preemphasis The rank order in overall speech quality can be seen from the projections on the hypothetical quality dimension, which goes at an oblique angle between the orthogonal dimensions. Splitting up speech quality into its components ob- intelligibility (at constant loudness) are viously gives more detailed information the best descriptors for the perceptual 3 Usability about the perceived difference between dimensions that are the fundamental After this overview of the concept of the systems. (In this kind of test the loud- components of speech quality. McGee speech quality and its components, it is ness has been equalised between the used semantic differential scales for appropriate to make a comparison with transmission bands, so that loudness is evaluating a large number of system con- usability concepts and look for possible not involved as a separate dimension in ditions. He found that the Naturalness relations between the two. Many differ- the evaluation.) factor was highly correlated with such ent definitions of usability are found in scales as Big–Small, Wide–Narrow, literature, some of which are language Figure 11 (from [1]) demonstrates how Full–Thin, Pleasing–Annoying, and Nat- specific and cannot be easily translated the effect of two different slopes of the ural–Unnatural. The Intelligibility factor into other languages. Furthermore, frequency response (either flat or with a was more related to Bright– Muffled, usability definitions are often opera- +6 dB/octave preemphasis) within the Sharp–Dull, etc. and, of course, to scores tional, i.e. what they mean depends on band is related to two extreme acoustic calculated from articulation index mod- how they are measured. conditions (reverberant or anechoic) at els. the send or receive end of a telephone In 1974 the Communications Studies connection. Again, naturalness and intel- The ultimate consequence of these find- Group, CSG, in London evaluated the ligibility have been rated separately. The ings is that the quality of speech systems usability of a few types of conference results indicate that naturalness is always under test can be rated directly on the telephones for a Scandinavian working better with a flat response but that intel- two dimensions of Naturalness and Intel- group. They performed a large number of ligibility increases somewhat in the two ligibility, for example on 5-grade scales laboratory tests with different groups of cases with reverberation at the send end going from “No Naturalness (or Intel- users, who had to use the various tele- when the preemphasis is applied. The test ligibility) at all” to “Complete Natural- phones for the same kind of tasks. Both results also prove that naturalness and ness (or Intelligibility)”. Figures 10 and the task performance and the attitudes of intelligibility are indeed orthogonal 11 are examples of actual applications. the users to the equipment were recorded. dimensions. By this method, the redundancy of rating scales in a questionnaire, and with that In the reports from these experiments, Other experiments, e.g. by McGee [5], the total test time, can be considerably CSG applied the concepts of Effective- have demonstrated that naturalness and reduced.

Telektronikk 3/4.1997 21 ness and Acceptability to the two prin- task. Typical examples of such dimen- The evaluation of user attitudes in a cipal dimensions for structuring the sions are Satisfaction, Acceptability, usability test is considerably facilitated results. By effectiveness they meant how Social presence, Comfort, Enjoyment, by identifying in advance those dimen- well the task was solved by the respec- Control, and Aesthetic appeal. sions of the acceptability/satisfaction tive equipment, and by acceptability how concept which have the best differential much the subjects liked using the equip- It should be observed that when usability power, i.e. which most clearly show the ment for the task. is measured, the outcome is always spe- difference between systems under test. cific for the kind of user involved, the Opinion rating scales which give the These two main dimensions are still used task to be solved, and the context and same score for systems that are perceived for both the description and the evalua- environment in which the test has been as significantly different are not effective tion of usability. Definitions along these carried out. ISO has furthermore intro- as test tools. In the same way, rating lines have been applied in the European duced the concept of Context of use, scales which all give the same score for a research project COST 212 (Human Fac- defined as: specific test object are probably redun- tors in Information Services) as well as dant and therefore less efficient. In this within the European standards institute of Context of use: The users, goals, tasks, respect the methodological situation is ETSI. Internationally, similar although equipment (hardware, the same as for the evaluation of speech not identical definitions are found in ISO software and materials), quality. 9241-11 “Guidance on usability” (1996). and the physical and social environment in For person-to-person telecommunication, 3.1 Usability concepts which a product is used. the Communications Studies Group have shown that the concept of acceptability Consequently, the concept of Usability comprises an important dimension which With the above background, measures of cannot be generalised unless it has been may be called Social Presence. This usability may be assumed to be of two verified how much it is affected by varia- factor is of major importance for the kinds: tions in the parameters user, task and comparison of different communication environment. Therefore, the additional systems and is therefore a suitable basis a) Performance measures, which are concept of Flexibility has been intro- for the design of questionnaires. Systems “objective” measures or observations of duced as a measure of these effects, in having a high degree of Social Presence user behaviour and are focused on task the sense that the less usability changes are judged as being “warm, personal, performance, i.e. how well the users can over variations of any of the three param- sensitive and sociable”. achieve a specific task. eters, the larger is the flexibility of the system in question. Another factor of particular importance b) Attitude measures, which are “subjec- for the evaluation of terminal equipment tive” measures of the users’ opinions of Finally, the change in Usability as mea- has been called Aesthetic Appeal. This working with the system, i.e. how much sured over a number of trials with test dimension is positively correlated with they like using the system. subjects representing users, is a measure adjectives such as colourful, large, of Learnability. Requirements for learn- spacious, beautiful and interesting. How- Performance measures include both ability need not be the same for different ever, Aesthetic Appeal is also strongly Effectiveness (e.g. error rate) and Effi- user groups: for infrequent users, e.g. related to the Social Presence dimension ciency (e.g. performance time), which are users of public telephone terminals, it is and may even be confused with it. considered to be independent factors, in often important that the use is successful the sense that they can vary without nec- even at the first trial, while for office “Aesthetic appeal” does not necessarily essarily being correlated. The following equipment it might be better to aim at a refer to any aesthetic or artistic value but definitions are given in ISO 9241-11: high efficiency for professional users, rather to the appearance, the perception which may be achieved only after a rela- of shape and colour and even the tactile Effectiveness: The accuracy and com- tively long training period. impression of the equipment representing pleteness with which the human/machine interface. No doubt users achieve specified these are components which have a sig- goals. 3.2 Usability and telecommunications nificant influence on the users’ attitudes towards a telecommunication system, Efficiency: The resources expended particularly its “acceptability”. There are in relation to the accuracy The ISO 9241 definitions of usability in fact numerous examples of products and completeness with components are of a general nature, but (be they telephone sets or hearing pro- which users achieve the standard in fact addresses particularly tectors for work in noisy locations) goals. the ergonomical requirements for office where the user has preferred the type work with visual display terminals. For having the more elegant light-weight Attitude measures include concepts such telecommunication terminals similar per- design over other types with better per- as satisfaction and acceptability. The atti- formance criteria may be applied; how- formance but with a less appealing look tude measure is rather more complex ever, the evaluation of user attitudes may and heavier weight. These examples than performance – it should be seen as a need further consideration when usability bring out the fact that it is the users who conglomerate of underlying dimensions, concepts are applied to telecommunica- judge which product is most usable by which are all related to the user’s perception systems and services. their personal weightings of performance tion and experiences of the system when against satisfaction or acceptability, not it is used for solving a certain type of

22 Telektronikk 3/4.1997 any “objective” external judges with their expert opinion. USABILITY

Further dimensions which have proven useful for explaining the results from attitude questionnaires especially for the PERFORMANCE ATTITUDES use of conference telephone systems are, according to CSG (1974): - Convenience Acceptability Satisfaction - Privacy Effectiveness Efficiency - Control of conversation Aesthetic Comfort - Polarisation between participants appeal - Concentration - Time pressure. Personal Privacy presence Another set of dimensions for the assessment of satisfaction has been proposed by the Heinrich Hertz Institute in Berlin Enjoyment Control in a contribution to the ETSI work on usability evaluation. They found in tests on some telecommunications services, CHECKLIST DATA particularly multimedia information systems, that the following three factors can be used for describing satisfaction in a communication situation: - Effort versus Ease of use Physical characteristics Software characteristics - Agitation versus Relaxation Figure 12 Usability and its components. The model has three layers, where the bottom - Boredom versus Enjoyment. layer consists of the physical factors that have an impact on the performance and attitude components, which are mutually independent but may depend on the same under- The Effort/Ease-of-use dimension is pos- lying factors. “Checklist data” are recommended values for the bottom layer charac- itively correlated with such adjectives as teristics in order to achieve an acceptable level of usability easy, simple and restful, Agitation/ Relaxation with relaxed and calm, and Boredom/Enjoyment with entertaining, animating and enjoyable.

3.3 A conceptual model type of conference telephone which in a detailed information about the strong and usability test was proven to be effective weak points of the product. Similarly to the model of speech quality and efficient for solving tasks by presented in Figure 2, the components of telecommunications that otherwise would Comparing the conceptual model for usability are expected to depend on a require a personal meeting, at the same speech quality with the usability model, number of underlying factors, as illu- time had a low acceptability for the it appears that intelligibility could be said strated in Figure 12. The two dimensions users, because it was perceived as bulky to correspond to the performance dimen- of usability: performance and attitude, and uncomfortable to use. In the same sion, while naturalness is related to atti- should be considered as orthogonal, i.e. test, another type of telephone was pre- tude measures comprising satisfaction, as mutually independent factors. Instead ferred because it was considered to be acceptability, etc. Intelligibility can be of being interdependent they may both easy to use, without actually being a very assessed by means of “objective” tests, depend on the same underlying factors efficient tool for the intended task. where the performance of the users is but to a different degree. They may also observed from the outside. The propor- depend on totally different factors. These If it were possible to estimate the weight- tion of correctly understood words is a factors could be any characteristics of a ing coefficient that an individual user measure of goal achievement and may be product or service that might have an would attach to the usability components, considered as the “effectiveness” as influence on the usability components, a total usability measure could be ob- defined by ISO. The number of words including “soft” factors such as the dia- tained that in the example above might understood per minute is a measure of logue structure in a voice response turn out to be about the same for both message rate and may be said to be the service. telephones, in spite of the obvious differ- “efficiency”. However, message rate is ence in performance and user satisfac- hardly a relevant measure for character- The assumption of orthogonality means tion. For the same reason as for speech ising a telephone system, because this that a product or service can get a high quality, splitting up usability into rate is much more dependent on the score on one component and a low score relevant components facilitates the users’ behaviour in a task solving on the other component as the result of a evaluation of a product and gives situation than on the transmission charac- systematic evaluation. An example: one teristics of the system.

Telektronikk 3/4.1997 23 rapidly growing usage is possible and USAGE AVAILABILITY may make the service an immediate success. Eventually, if a comprehensive Reliability Regulations pool of empirical data is available (e.g. checklist data as shown in Figure 12) and the relations between the factors in a UTILITY Infrastructure structural model are reliably known, it may become possible to predict the COSTS NEEDS success of a new service before it is introduced. USABILITY 4 Service quality

Originally, “Quality of Service” dealt Performance Acceptability exclusively with transmission performance and reliability, particularly in Figure 13 A model of Usage, Utility and Usability ITU-T documentation. However, recently more and more attention has been devoted to perceived performance, which means that customer satisfaction is coming into focus. In ITU-T Recommenda- tion F.70 [10] the main criteria of service performance from the users’ viewpoint are: Naturalness, on the other hand, can only Utility may be seen as the relative value 1)ease in establishing a connection be assessed by subjective tests, where the a user attaches to a product by taking the users have to give their opinion on spe- costs into account together with the bene- 2)retention of the established connection cific rating scales. It is therefore compar- fit of using the product for a specific 3)satisfactory transmission quality able to the “satisfaction” component of purpose. usability. The difference is merely that 4)integrity of billing. while naturalness may be considered to One advantage of such a conceptual be one-dimensional, the attitude compo- model is that it clearly demonstrates the Applying a usability perspective to nents of usability consist of a rather large role of usability in relation to the cost service quality may be facilitated by number of independent dimensions and the user’s need. It is a necessary but adopting the classification of tasks which which all have to be rated separately, if a not sufficient condition that usability has been developed within the European complete evaluation of the usability of a reaches a certain minimum level. How- RACE/GUIDANCE project. A distinc- product is aimed at. ever, shortcomings in the level of tion is made there between goal tasks, usability may be compensated by either a which are the tasks that the user wants to Therefore, the principal difference strong need for using the product or by a perform, and the enabling tasks, which between the concept of speech quality low cost, or both. Maximum utility is are what the user must do to create a state and the even more complex concept of obtained by the highest possible usability which enables the goal task to be per- usability is that it is hardly meaningful to together with a sufficiently low price in formed. The first point in F.70 deals with try to estimate usability directly on a sin- relation to the needs of the user. With the enabling task, the second point with gle scale in a test, while it is often possi- this model, utility comes down to zero reliability, and the third point with the ble to obtain a relevant overall speech even for highly usable products, if the goal task. The fourth point is a different quality measure without going into fur- consumer does not need them, or if the issue, which today might be generalised ther detail. price is out of reach. to deal with service integrity and security.

3.4 Utility and usage With a given utility of a product for a rel- For a telephone service, usability aspects evant user group, there will be a certain can be applied both to the access phase demand which finally leads to some (enabling) and to the speech communi- It should always be kept in mind that degree of usage of the product. However, cation phase (goal) which generally make usability is a quality measure that does there are further conditions which have up the service. For both phases, usability not include the costs for providing the an influence on the usage. One group of targets can be set up and evaluated. The product or service. It relates only to the such additional factors may be called ease in establishing the connection is ergonomic quality of using the product. Availability. Especially within the area very much dependent on the procedure Even if “efficiency” involves the re- of telecommunications, it is a necessary for setting up the call, which may involve sources expended, which includes the condition to have access to a reliable net- access codes, verbal dialogues with an physical and mental effort as well as the work and to be able to reach a sufficient operator or a speech recogniser, payment time spent by the user for task comple- number of other users of the same by coins or cards, etc. User instructions tion, it should not include the financial service. If all conditions are fulfilled also play an important role for the ease of costs for providing or using the product. when a new service is introduced and the access, as well as the design of the physi- Instead, the concept of Utility should be usability and utility is high enough, a cal interface in the terminal. applied, as illustrated by Figure 13.

24 Telektronikk 3/4.1997 Therefore, the assessment of service 5 Conclusions 2 Gleiss, N. Sound transmission qua- quality should always involve a usability lity : the influence of signal-to-noise evaluation of the complete access phase It has been shown here that the concepts ratio and bandwidth. Tele Engl. Ed., of the service by measuring user satisfac- of speech quality and usability both stand 2, 1972. tion as well as performance (e.g. per- on a similar basis and can be described centage of achievement at first trial, or by parallel models. Speech quality has 3 Gleiss, N. Preferred listening levels error rate, on the one hand and time for two basic components: one is related to in telephony. Tele Engl. Ed., 2, 1974. completion on the other hand). speech as a means of communication and describes the performance of a telephone 4 Gleiss, N, Goldstein, M. Loudness of The communication phase should be system when transmitting semantic infor- speech for monaural and binaural lis- dealt with separately. For evaluation of mation, and the other is related to con- tening. 7th FASE Symposium, Edin- the quality of communication by speech, veying personal and emotional character- burgh, 1988. the division of speech quality into per- istics imbedded in the voice of the ceptual factors, which may be studied speaker. Similarly, the usability of a ser- 5 McGee. Semantic components of independently, is a useful approach. vice has an “objective” dimension, de- processed speech. Journal of Acoust. Presently, there is a clear tendency to scribing goal achievement and efficiency, Soc. Am., 1964. apply the task oriented methodology and a “subjective” dimension, describing implied by the usability model also for user satisfaction and acceptability. 6 ISO. Ergonomic requirements for the communication phase. office work with visual display termi- Both models can be integrated in the nals, Part 11 : Guidance on usability. As an example, according to ITU-T Rec- evaluation of service quality. They are ISO 9241, 1996. ommendation P.85, the quality of a applicable both with respect to different speech synthesis system should be evaluation mehtods and for the assess- 7 ETSI. Speech communication quality assessed by application of the system in a ment of service quality by calculation from mouth to ear for 3.1 kHz hand- simulated service where the subjects per- methods. The models may also accom- set telephony across networks. form specific tasks in a given scenario, modate additional factors such as ETSI/ETR 250, 1996. e.g. a railway traffic information service. reliability and availability as well as user In the test, the performance of the sub- expectations in the context of use and, 8 ITU-T. A method for subjective per- jects is recorded, and the degree of ultimately, also the cost of using the ser- formance assessment of the quality of acceptance of the voice service is vice in relation to the benefit for the user. speech from voice output devices. assessed by asking a question like “Do ITU-T Recommendation P.85. you think that this voice could be used Descriptive models based on carefully Geneva, 1994. for such an information service by tele- defined fundamental concepts are a good phone?” After that, more specific speech help for understanding why some 9 ITU-T. Transmission impairments. quality rating scales are applied, such as: telecommunications services succeed and ITU-T Recommendation G.113. Overall impression, Listening effort, others fail. They may furthermore be Geneva, 1996. Articulation, Pronunciation, Pleasant- considered as useful tools for research ness. The main point is that the synthesis and development which can promote the 10 ITU-T (CCITT). Evaluating the system is not rated per se but in a service design of both effective and more user quality of the international telex ser- application. In this way speech quality oriented services. vice. ITU-T Recommendation F.70. can be made an integral part of service Geneva, 1988. quality. References 11 ITU-T. Models for predicting trans- A further factor that may affect perceived mission quality from objective mea- 1 Gleiss, N. Desirable sending fre- service quality – apart from all the surements. Supplement 3 to ITU-T quency response of telephone sets. factors contained in the usability concept Series P Recommendations. Geneva, Tele Engl. Ed., 1, 1989. – is the users’ expectations. A user’s 1994. opinion of a service is no doubt coloured by expectations about what can or cannot be achieved by it and how the possible benefits compare with other services. For example, it appears that users of mobile telephones have a more lenient attitude to the speech quality of mobile services in comparison with the fixed telephone net- Norman Gleiss is the former Human Factors Manager work services, because the advantage of (now retired) at Telia Research, Sweden. He is also the former Chairman of ETSI Technical Committee Human mobility makes the user judge a degrada- Factors. tion in sound quality less seriously. This factor has been taken into account in a e-mail: model for the calculation of speech [email protected] quality recently developed by ETSI [7].

Telektronikk 3/4.1997 25 Some current issues of Human Factors in telecommunications KNUT NORDBY

Scope of the review exchanges came into use. This means length of time, etc. The manufacturers that three to four generations have grown maintain that it is not worth putting any Human factors in telecommunications up with and learned this procedure. As a resources or efforts into the ergonomics covers such a large and diverse field that result of the emergence of new techno- of telephones, which are looked upon as it is quite impossible to cover in any logies, with the merging of telecommuni- cheap consumables rather than invest- detail all its varied aspects in a review cation and information technologies, and ments. It is hard to see the logic of imple- like this. However, I will here try to give the competition made possible by de- menting many features which most users an overview of some current topics for regulation and privatisation of operating will never use, except as a marketing those who are new to the field, but companies, a multitude of new services ploy. Badly designed telephones, how- readers who wish a more thorough treat- have grown up requiring new interfaces. ever cheap they may be, will always be a ment must look to the specialist litera- Most notable are the interactive voice liability to the users, and may ultimately ture. The most redeeming feature of this telephone services where you are asked lead to missed business opportunities, overview is probably that it is reasonably to press number keys to input your which no company can afford. When up to date since much of it is based on choices, or with voice recognition, where new telephones were introduced at the contributions to the 16th International you may speak your choices. Since most University of Oslo, hardly any users Symposium on Human Factors in users have grown accustomed to interact could perform even the simplest actions Telecommunications (HFT’97) held in with people via the telephone, interacting like call transfer, but had to ask the caller Oslo, Norway, 12–16 May, 1997 [1]. with the impersonal telephone system to call back to the operator and ask for itself is a new experience to most people the other party. This was not merely a This review is meant to present an and has to be learnt by the common user. learning problem, but the result of in- account of what is currently going on Unfortunately, the design of these new different interface design. Most staff just within the many fields of human factors interfaces has been aimed more at well would not take the effort to look up the in telecommunications research in vari- educated, computer literate, young men, procedure in the instruction manual every ous laboratories around the world. To than at common users, including people time they had to transfer a call, and the simplify the overview, the different top- of all ages, all education levels and all operation was too complex, involving ics have been grouped together under levels of mastering. arbitrary codes, to remember if you did eight broad headings, corresponding not perform the operation regularly. roughly to the main sessions of the The oldest and most important form for Also, there was no feedback from the HFT’97 symposium. These are: feedback to users of the telephone system system to inform you if you had per- is the line signals or ‘tones’ you hear in formed the operation correctly, often 1. User interfaces – simple and complex the handset. Most users have now grown leaving the unfortunate callers on hold 2. Speech interfaces accustomed to their ‘local tones’, but indefinitely. recently a number of new tones have 3. Video communication and multimedia appeared, confusing the users. If you 1.1 Audible tones in public tele- 4. Design for all – support for elderly and travel abroad or phone another country, phone networks disabled you will often encounter new and strange tones, whose meanings are not always 5. Group working support services The most important feedback to the users obvious, leading to more confusion and of the public telephone system is the var- 6. Product development uncertainty. There is obviously a strong ious tones you hear in the handset (i.e. need for standardisation and harmonisa- 7. New services, user requirements and the line signals) such as the dialling tone, tion, but should we keep the ‘hard core’ marketing the ringing tone, the engaged tone and of traditional tones or should we look at the multitude of other more or less in- 8. Convergence – integration of tele- the information tones from a different telligible tones introduced by various phony, computing, Internet. angle and design new tones which con- operators. vey the meaning and degree of urgency In a review like this it is difficult to avoid in a more universal way. Traditions are A comprehensive study of the human technical terms. However, I will try to hard to change and many people are loath factors aspects of information tones in keep the use of very special technical to give up their well-known tones to public telephone networks was carried terms and abbreviations to a minimum some strange sounds imposed upon them out by D. Anderson, A. Ferris and W. and to explain them as they appear. A by a distant and impersonal international Mellors, which is published as ETSI full list of references is provided at the agency. Report ER 145 Part 1 and 2 [2,3] (ETSI end of the review, but those who want to = European Telecommunication Stan- go into any depth are advised to consult It is remarkable that so many new dardisation Institute). The work was the specialist technical literature. features are currently implemented in mandated by the CEC (Commission of new telephone instruments without the the European Communities) to investi- necessary attention to human factors of gate the current and future use of infor- 1 User interfaces the user interface. New functions abound mation tones in the public telephone net- with buttons marked with non-standard, – simple and complex work with the view of harmonisation in unintelligible symbols, new user pro- Europe to avoid the possibility of cross The telephone user interface has re- cedures breaking with established prac- border misinterpretation and confusion of mained nearly unchanged for a very long tices, bad ergonomics with no tactile tones by users from different countries. time, basic call set-up procedure being feedback from keys, illegible symbols There is an urgent need for harmonisa- virtually the same today as 80–90 years and displays, and badly designed hand- tion and simplification since an initial ago when Strowger automatic telephone sets which are strenuous to hold for any

26 Telektronikk 3/4.1997 survey by Gagliardi had found that at least 228 different information tones Talk Phase were in concurrent use in Europe alone! (see [2,3,4,5,6]; and companion article by Anderson & Ferris, Feedback tones in public telephone networks: Human Fac- Interaction tors work at ETSI, in this issue of Telek- with System tronikk). Anderson, Ferris & Mellors have de- Progress Failure to Prompt for veloped a three-dimensional user-goal model (Figure 1) where each axis repre- Progress Action sents a semantic pair: Basic Basic 1. Progress – Failure to Progress

2. Prompt for Action – No Action Re- Sophisticated Sophisticated quired 3. Interaction with System – Talk Phase.

Based on this model and functional de- Progress Failure to No Action finitions of the most basic line signals, Progress Anderson et al. devised some new tones Basic Basic Required to match the various semantic characteristics defined in the model: such as the degree of urgency, the phase of the call, and call progress. It is thus possible to Sophisticated Sophisticated propose basic information tones with high human factors relevance to replace today’s tones which were based on tech- Figure 1 Three-dimensional user-goal model for categorising network information nical possibilities in the early telephone tones in public telephone networks along three dichotomy dimensions. These are: networks rather than on human factors. Progress ↔ Failure to progress; No action required ↔ Prompt for action; and System interaction phase ↔ Talk phase. Both existing and proposed information tones can be Each new tone, as well as some existing characterised along these three dimensions (after [6]) commonly used tones, have been subjected to an ‘audit’ based on the functional definitions, best fit with the user- goal model, and on technical characteristics. As a result if this audit, Anderson et questions. Initially, Helmreich observes right, providing it doesn’t cost any al. propose six new tones for Pan-Euro- that to sell a product one must advertise extra”. He then goes on to show that the pean standardisation: dial tone, ring tone, its many functions. The customers must use of ergonomics in telephone design busy tone, congestion tone, special infor- believe that the more functions a product does, in fact, pay off in real money. He mation tone, and pay tone [2,3,6]. The possesses, the greater benefits will be compares a set of newly designed tele- tones have not been tried on common derived from buying it. Common ex- phones from Siemens, where the best users yet, but knowing the users’ general perience, however, has shown that most ergonomic principles have been used, to resistance to any change, their reactions novel features are rarely, if ever, used – some conventional telephones, and is to the new tones should be interesting. over-featuring seems to go hand-in-hand able to show in a series of analyses that with under-use. the user-friendly telephones do, indeed, 1.2 Is it worth having ergonomic give an economic benefit over the con- telephone user interfaces? Sadly, Helmreich points out, the ergo- ventional telephones. nomic aspects of many telephones in A recent problem for many users is the widespread use do not even meet mini- Helmreich first looks at the learning many functions found in modern tele- mum user requirements. Low contrast phase, where he can show a saving dur- phones, which no-one seems to use. Why displays are difficult to read, very minute ing the introductory learning phase of are user interfaces so poor; why have the buttons and dialling keys are often between DEM 12.50 and DEM 31.25 per human factors specialists had so little ‘spongy’ (i.e. have no clear tactile feed- person. Everyday use, however, shows impact on product development; and it is back), operating procedures are incon- even greater savings. Helmreich shows really worth the cost and effort to sistent and markings on operating con- that in an office with a PABX with one improve the user interface of telephone trols are either incomprehensible, thousand lines the savings from the use functions no-one wants to use. illegible, or both. of the new telephones (especially in increased use of call transfer and con- In a critical paper, Helmreich [7] from Helmreich further maintains that the ference calls) add up to 75 seconds per Siemens, Germany, focuses on these most deciding factor for telephone manu- week, giving an overall saving of up to facturers are costs: “Ergonomics is all DEM 206 per week for each telephone.

Telektronikk 3/4.1997 27 scheduling appointments can be con- 1996 siderably simplified or even fully auto- Sun Mon Tue Wed Thu Fra Sat mated. Dieterich & Steiner [8] of 1 2 3 4 5 6 7 8 9 10 11 Siemens, Germany, show that 12 13 14 15 16 17 18 autonomous ‘intelligent agents’, acting 1996 1996 19 20 21 22 23 24 25 on behalf of their users, can take over the • • 26 27 28 29 30 31 • • tedious task of negotiating appointments. • • • • Intelligent service nodes in telecommuni- • • • • cation networks will not only enable • • agents to communicate with each other, but also to dispatch messages via fax or e-mail, or convert them into interactive Inter- voice messages sent to telephones or action Personal Information Communicators. Inter- Commu- Network Commu- action A nication nication A The basic idea put forward by Dieterich & Steiner is to utilise existing infrastructure (telephone and GSM networks) for easy accessibility, widely available CalendarTool A CalendarTool B electronic calendar tools for wide availability, and a soon to be standardised open communication and negotiation protocol [9] (see Figure 2). P P ersonal Data/A ersonal Data/B Dieterich & Steiner used the Multi-Agent Interaction and Implementation Lan- guage (MAI2L®) developed by Siemens Figure 2 The architecture of the appointment scheduling system: The users work with for implementing intelligent agents in calendars they are used to. Intelligent agents, acting as wrappers, communicate with C® on Unix® and Windows32® plat- each other via standardised protocols, which allows integration of multiple calendar forms as part of the Multi-Agent En- systems (after [8]) vironment for Constructing Co-operative Applications (MECCA®). The scheduling capabilities of such agents are still limited to finding a time slot that is free for all. It will not consider constraints Helmreich also looked at factors that development process, he believes that such as the time it takes to travel to the cannot be quantified. By having tele- there will hardly be any increase in meeting venue or various personal priori- phones that are easier to use, the running development costs [7]. ties. However, Dieterich & Steiner think of the daily business will be more effi- that a wide variety of techniques de- cient. The telephone has become such an 1.3 Intelligent agents as veloped for distributed appointment important instrument in the daily running appointment managers scheduling, such as automated learning of businesses that mistakes or lack of of user preferences, priority management confidence in e.g. transferring a call, can, and conflict resolution, can easily be Time management has always been a in the worst case, lead to loss of cus- incorporated into this framework. Stan- problem, both in work and leisure. A tomers and revenue. Such factors cannot dard human-computer interaction tech- variety of support tools for time planning easily be quantified, but may have a con- niques, such as natural language under- and management have been devised in siderable effect on operating results and standing, should also be of use for recent years, first paper based, but in- productivity. human-agent interaction. creasingly in electronic form. The task of scheduling an appointment is both time The savings found may not look impres- consuming and requires a high degree of sive, but an average of DEM 200 a year 2 Speech interfaces co-ordination; especially when more than per employee, however, will quickly add two participants are involved; each of The most direct human information inter- up to savings running into many tens of which may have a busy schedule. Tradi- face has always been speech, which is a thousands of marks a year in larger com- tionally, several iterations, face-to-face, genuinely human ability acquired by vir- panies. Paying, say as much as via telephone, fax or e-mail were neces- tually all the members of the community. DEM 100, extra for a more ergonomic sary to find a time acceptable to all Until recently, however, the technology telephone will, thus, pay for itself in only parties. The likelihood that not all partici- necessary for consistent recognition of six months. In conclusion, Helmreich pants are on hand at the same time com- the human voice has not been available, maintains that if it is possible to demon- plicates further the joint scheduling pro- and alternative means for interaction with strate economic benefits for manu- cess. telephone systems have been developed. facturer and customer, there is a greater With the introduction of DTMF (= Dual chance that ergonomic design will be With the increasing use of electronic Tone Modulation Frequencies), ‘touch- used when developing a new product. If time management tools, the process of tone’ to some people, it became possible this is done as an integral part of the

28 Telektronikk 3/4.1997 to utilise the telephone for other things ‘confirm’ choices and as ‘delimiter’ key here to stay, but we have to design its than conversations or one-way informa- to terminate a service. Even the standards dialogue properly for it to survive in the tion services (e.g. the speaking clock, published by ETSI (European Telecom- marketplace and make revenue. news, weather reports, etc.). By being munication Standards Institute) and the able to give input to the system, users recommendations from ITU-T (Inter- The following sections illustrate various could engage in interactive services with- national Telecommunication Union) attempts to design interactive telecom- out the intervention of a human operator. show inconsistencies, and harmonised munication services and to compare dif- One of the first interactive services to be standardisation is urgently needed. ferent dialogue designs, both in the labo- offered to the general public was a ser- ratory and in real-life settings. vice allowing people to check their own The alternative to using the dialling key- bank accounts. This is now ‘tele-bank- pad is to use voice recognition allowing 2.1 A common keypad interface ing’, allowing people to transfer money the user to speak the various commands for voice-response services from one account to another or pay bills and inputs to the system. The problems via their home telephone without the associated with making a good and Different telephone voice-response ser- need of any extra equipment. Another simple user interface with this technol- vices utilising the dialling keypad (e.g. early service was provided by book ogy is even more daunting and much messaging, interaction and information clubs, allowing their members to cancel effort has gone into this during the last retrieval) employ different functional or order books via the dialling keypad of years, as demonstrated below. interfaces, leading to users having diffi- their home telephone. culties when switching between services. The inherent impreciseness of colloquial Goldstein [10] of Telia Research, Today, such services have become so speech makes it difficult to design a dia- Sweden describes a common functional common that it is hardly possible to logue that will work smoothly for differ- keypad interface proposed by Telia Re- phone any agency or large business con- ent users in different contexts: It reminds search, and how it was evaluated in three cern without being greeted by: “If you me of the Norwegian folk tale about a voice-response services. want this, press 1, if you want that, press deaf man who, seeing the sheriff 2”. For most people the procedures for approaching, carefully thinks out clever By using the ‘Tape Recorder’ metaphor using many of these systems are too answers to the questions he expects, and together with the auditory AlphaWheel complicated since you have to remember the hilarious dialogue which follows paradigm and the ‘Constrained Output service codes, account numbers, item when the questions do not match his Set’, a universal navigation concept, codes and key-mappings, leading to replies. Speaking to a telephone system called a common functional keypad inter- ‘slam-down’ and under-use. still puts many people off, especially if face, was developed by Telia for use with the dialogue is jerky and unnatural. a voice-messaging service (MobilSvar). The handling of error recovery in most Speech recognition as input to interactive The basic question Goldstein asked was services leaves much to be desired – telecommunication services is surely most people who make an error just hang up. Feedback is another neglected issue. Even minute differences in the wording of prompts and feedback messages can mean the difference between success and 1 2 3 ruin of a service. In an interactive service introduced in UK, the users were asked to “Press the harsh key”, which lead to Menu users not understanding how to proceed. options Although the term hash is the ‘correct’ name for the # symbol, it is generally 4 5 6 known as the square in UK, and when Change/Erase the message was changed, the service was a success. Inconsistent naming of the two symbols leads to confusion. The * Tape symbol is usually called asterisk or star, 7 8 9 recorder or their equivalents in different langu- metaphor ages, but the # symbol has many differ- Backwards Start/Stop Forwards ent names, even within one language, e.g. harsh, square (UK), pound (US), number sign, sharp (music), double Decision cross, little gate, fence, to mention only a * 0 # functions few. More important, though, is the Undo/Cancel Help Finish/Execute inconsistent use of the two symbols. The * key is usually used to ‘initiate’ a service, as ‘separator’ between different Figure 3 A common functional keypad interface proposed by Telia for all kinds of inputs or as ‘affirmative’ key, but also, interactive voice response services: Keys 1–5 for menu options; key 6 = change/erase; and inconsistently, as ‘delete’ keys 7–9 for the ‘tape-recorder’ metaphor (key 7 = fast rewind, key 8 = stop/start, (backspace) key, while the # key has key 9 = fast forward); and keys * 0 # for decision functions been used both to ‘initiate’ services, to (* = undo/cancel, 0 = help, and # = finish/execute) (after [10])

Telektronikk 3/4.1997 29 if this interface could also be used in an mentation. However, if flexibility and England, performed two experiments; information retrieval service. learnability are also taken into account, one on security data types and dialogue the results are in favour of Telia’s func- messages, and one to determine ‘insult The Telia functional keypad interface tional interface since no code manuals rate’ and verification completion. The conceptually divides the standard tele- are required. By offering customers one experiments addressed the question of phone dialling keypad into three areas common functional interface for all inter- how to proceed in dialogues with front- (see Figure 3): Menu options (1 2 3 4 5), active voice services, transfer from one end speaker verification to design ‘best Change/erase (6), the ‘Tape Recorder’ service to another is made easier. The practice’ hand-over strategies for trans- metaphor (7 8 9) and Decision Functions functional keypad interface is cheap to ferring a user to a human operator should (* 0 #). implement both from a user’s and opera- difficulties arise. tor’s point of view, as it is based on The ‘Tape Recorder’ metaphor involves available technology and is terminal and The results showed a general (and ex- mapping the keys of the standard music language independent. It is also sup- pected) decline in user attitude with in- cassette player onto the 7 8 and 9 keys ported by mobile telephones. creasing ‘insult rate’. Although there was so that the 8 key plays the message, the 7 no significant difference between suc- key goes back (rewind) and the 9 key The proposed common functional keypad cessful completion by machine or human goes forward (fast forward). Functions interface does not conform to current operator on the second attempt, analysis are toggled on and off by repeated key ITU-T or ETSI standards for definitions of individual statements point to a prefer- presses. of telephone keypad key functions. More ence for the second attempt being carried human factors assessment and extensive out by machine rather than suggesting an The auditory AlphaWheel paradigm is a full-scale user trials will be needed immediate hand-over to a human opera- software application for entering alpha- before standardisation of the functional tor. These results may be explained by a betic input in any language from a stan- keyboard interface can be undertaken. different voice being introduced by the dard telephone numeric keypad. The sys- human operator. This requires heightened tem recites the letters of the alphabet “A, 2.2 Speaker verification secu- concentration, increasing the cognitive B, C, ...” etc. and when the first letter of rity in automated services load. the desired name has been reached the user selects (or ‘clicks’) it by pressing the First and foremost, their findings show Procedures for security checks in auto- # (‘Execute’) key. The system (if neces- that users have strong opinions towards mated telephone services, such as home sary) recites the alphabet again and the these issues and, consequently, that the banking or home shopping, require par- second (and third) letter of the desired speaker verification procedures in auto- ticularly careful design of the dialogue name is selected. By using the ‘Tape mated telephone services need to be care- between system and user. This is because Recorder’ metaphor keys, the user can fully researched and designed. of the ‘binary’ nature of the access deci- move quickly through the alphabet (9 sion; you are either accepted or rejected. ‘Forward’ +5 letters and 7 ‘Rewind’ –5 There are two options for the designer, 2.3 Experience with ‘Voice letters at a time). When enough letters either to use user-specific knowledge, Server’ for overseas calls (usually two or three) have been selected, such as passwords and PIN-codes, or to the ‘Constrained Output Set’ will recite a use biometric technologies, which rely A voice recognition ‘Long Distance short (3-5) list of names beginning with on specific physical attributes of an Voice Phone Calls Server’ for managing the selected letters. The subject can then authorised user, such as speaker recogni- overseas telephone calls for callers who choose the desired name by ‘clicking’ it tion, fingerprints or iris identification. had the necessary information and had no with the # key (if there is only one name Of these, automatic speaker verification need for other services had been intro- in the list beginning with the first selected holds distinct potential since security duced in Colombia in 1995 as a comple- letter it will be selected immediately). would not only depend on knowledge of ment to telephone operators. Charry, passwords or digit codes, but also on Zuñiga & Mejia [12] of ITEC-Telecom In the user trials a Telia bank-by-phone user-specific voice attributes. Unlike R&D, Bogotá, Colombia, studied the service, using the common functional ordinary dialogue design, where errors user reactions and user acceptance of this keypad, was significantly more efficient can be dealt with by well-designed error system. than a comparable service (Datasvar) on recovery strategies, verification of user the Swedish market. Two public infor- authorisation results in either access or Due to the diversity of Spanish dialects mation retrieval services (a Ferry Service rejection. In the case of rejection, system spoken in Colombia, the varying levels with 210 port names, and the Greater designers can either provide immediate of technical training, and the need to Stockholm Public Transport Service with human-operator backup, or introduce speak telephone numbers one digit at a 350 station names), both which use a repeated automatic verification cycles, time, the system was not very well re- digit code paradigm, where implemented with the risk of increased ‘insult rate’ ceived. Initially in the Bogotá region, with Telia’s common functional keypad should the user fail to gain access after 45 % of the callers hung up when interface. Comparisons between the orig- repeated verification sequences, or termi- encountering the new service, 20 % inal services and Telia’s implementations nate the transaction. switched to a human operator, 16.5 % gave diverging results; the Telia ferry failed to complete the call and gave up, service was significantly more efficient Schmidt*, Ironside*, Jack* & Stentiford# and only 18 % completed successfully than the code-dependent service, while [11], *Centre of Communication Inter- the attempted calls using the Voice the original code-dependent Stockholm face Research, University of Edinburgh, Server (0.5 % of the calls could not be Public Transport Service was signifi- Scotland, and #BT Laboratories, Ipswich, completed for various technical reasons). cantly more efficient than Telia’s imple-

30 Telektronikk 3/4.1997 Even after having been in use for some time and after changes and improvements to the system, these numbers did not Speech change dramatically: (38 %, 38 %, 8 % recognition and 15.5 %, respectively). Actually, the number of successfully completed calls fell from 18 % to 15.5 %, but more people Wizard et switched to a human operator. Charry `recognition` al. propose various cultural, linguistic and ergonomic explanations to this: SDL/SD T Telephone dialogue spec. interface • The system was introduced abruptly and there was a lack of publicity. Digitised speech • Colombians do not feel any need for using such technology and are even afraid of using it, although they think that the company does not provide suf- Speech ficient operators to provide proper ser- synthesis vice. • The majority of users had never used any type of voice recognition technol- Figure 4 The test environment includes a tool where dialogues may be specified in SDL (Specification ogy before and therefore rejected the and Description Language). This tool contains a runtime module that allows specifications to be com- service. piled and run in real-time in the same environment. This allows rapid prototyping and high consistency • The demand to speak long telephone from specification to implementation (after [13]) numbers one digit at a time was quite unnatural to most people, and callers had to write numbers down to be able to read them out in this manner. • Colombians have strong feelings of regional belonging, and not being and voice recognition is improving in synthesis and digitised speech and be- greeted in their local dialect put many most major and even some small langu- tween speech recognition and the Wizard off using the system. ages. In cultures where high tech is of Oz simulation of recognition (see Fig- recognised, resistance to new services ure 4). Traditional tools and methods, • Spanish synthetic speech is not yet will abate. But this does not give opera- such as Wizard of Oz simulation, SDL well developed, and many users found tors the freedom to introduce any bad (Specification and Description Lan- the prosody strange and rejected the service. The demand for high quality guage) and MUSE (Method for Usability system. interactive services will rise, and as com- Engineering) were utilised (Wizard of Oz • Spanish voice recognition is in an petition will give a choice between simi- = covert simulation of speech recognition early stage and has a very limited lar services, users will choose those and synthetic speech). vocabulary. which are simple to use. Hamnes et al. first set up some user • The interface did not provide sufficient requirements and design guidelines: help and information to instruct callers 2.4 Dialogue design for auto- in using the system correctly. matic speech recognition in • The successful transaction of sending a number paging number to a pager should not take • Any additional help, however, is too noticeably longer to complete than the time consuming, frustrating the callers’ To send a message (number) using an same transaction using an operator- patience. automatic numeric paging service re- based service. • The system is not ‘dynamic’ enough; quires tone signalling (DTMF) and tele- • The user should not require any train- many speakers interact like in a con- phones with * and # keys, but if this is ing other than a short instruction versation, starting to speak before they not available an operator-based system is printed in the telephone directory hear the prompt tone. required. Operator-based numeric paging (equivalent to the current automatic or incurs high costs on the service provider. • It is not possible to recover from an operator-based services), and the dia- The approach to design an automatic error, and you have to hang up and try logue should be easy to learn. number paging service for users with no again. access to DTMF, to reduce the use of • ‘User friendliness’ should be equiva- operator-based services, is described by lent to or better than the operator- For the Voice Server to be viable, Charry Hamnes, Amdal, Kulseth & Rugelbak based service. et al. feel that it is important that the dia- [13] from Telenor Research and De- logue is natural, free flowing and that • The service should issue an initial velopment, Kjeller, Norway. The work regional dialects are respected. Some of ‘welcome’ message, telling the user was carried out in a speech-based dia- the problems pointed out will be solved which service has been reached. logue experimentation environment, in the near future. Also, voice synthesis allowing easy switching between speech

Telektronikk 3/4.1997 31 • The service should clearly differentiate the operator-based service. Dialogues A cess. The analysis supported hypothe- between the pager number and the and B both had slightly longer comple- sising about the usability of certain com- message number being sent to the tion times. This had been expected since binations of dialogue parameter values. pager. they both apply isolated digit recogni- Due to low number of subjects, the val- tion, which leads to slower number input. idity of their quantitative results may be • The service should give feedback of Dialogue D (connected digit recognition, low, but findings were fairly consistent both the pager number and the mes- user-specified message length and a ver- and supported hypotheses made prior to sage. At some point the whole number bose style) gave a significantly longer the study. ‘Wizard errors’ (i.e. subjects (pager or message) should be read completion time. This was partly due to detecting the deception) were low and back to the user, so that the user may subjects not using the ‘barge-in’ facility, their effect should be minimal. either confirm or cancel if the number which would have reduced the task com- has been recognised incorrectly. pletion time. 2.5 Performance, preference • The service should indicate to the user and design of speech that the message has been sent and All four dialogues showed clear learning recognition systems give a clear exit message (observing effects. Learning time was determined by Ciganek’s tenet of closedness, i.e. that comparing the actual time used with a Even though the technology now exists a transaction should have a beginning, reference time measured for each of the for implementing applications using a task stage, and a defined end). four dialogue styles. For dialogues A, B speech recognition, we should not and C, the task completion time was very • The dialogue must be able to handle assume that speech recognition interfaces close to the reference times by the time error situations, e.g. the user giving no necessarily would be preferred over of call number three. The task comple- input or the wrong input. touch-tone or other interface types by all tion time with dialogue D was still signif- users. In a previous study, Fay [14] had icantly longer than the reference time by Hamnes et al. identified and analysed found that users preferred touch-tone the third call, which may be due to some dialogue parameters such as: (DTMF) in a college registration service, callers not using the ‘barge-in’ facility. but speech recognition in a telephone dis- • Speech recognition technologies connect service. All four automatic dialogues were rated • Grouping of digits for feedback higher than the existing operator-based D. Karis of GTE Laboratories, USA, • Prompts service, particularly with respect to qual- studied two different services to get data ity of feedback. Dialogues A, B and C • ‘Barge-in’ on user preferences of speech recognition were rated higher with respect to overall vs. touch-tone, user ratings and accept- • Dialogue ‘style’. ‘user friendliness’. Users preferred ance of a system as a function of the immediate feedback on each digit rather recognition performance experienced, On the basis of this, four dialogue pro- than feedback to groups of digits. The and general and specific user interface files were chosen for prototyping by operator-based service, however, was design techniques [15]. He also add- assigning values to the different dialogue rated higher with respect to subjective ressed the design of speech-only inter- parameters (see Table 1). A Wizard of Oz feeling of time used, with the exception faces as well as dual-mode (speech & simulation study was then carried out to of dialogue C, which used the most touch-tone input) interfaces. evaluate the four dialogues with respect sophisticated recognition technology. to the user requirements, and to compare Data was gathered while developing two them with the existing operator-based In conclusion, Hamnes et al. observe that GTE services: Speech Connect®, an service. by systematic approach based on the automated business directory service MUSE method, application of guidelines where callers can speak the name of the The findings showed that only dialogue and analysis of dialogue parameters, business they want to reach, and InCon- C (connected digit recognition, terse many design alternatives could be ruled nect®, a call management service. In style) had a shorter completion time than out at an early stage of the design pro-

Table 1 Dialogue profiles

Profile ASR Grouping Establishing Grouping Digit Barge-in Dialoge name techn. of pager # msg length of msg prompt style

A isolated 3-2-3 ‘end’-command none beep no terse

B isolated single digit user-spec. length single digit feedback yes verbose on digit

C connected whole number ‘end’-command none none yes terse

D connected 3-2-3 user-spec. length automatic none yes verbose

32 Telektronikk 3/4.1997 Speech Connect® the active vocabulary In conclusion, Karis proposes the follow- P:“Have you had any previous experi- is over a thousand words at some points, ing five specific design rules: ence of work in this area?” the system is speaker independent, does 1. Provide brief initial prompts followed, C:“No, but I have always wanted to work not support ‘barge-in’ and there is no if necessary, by more detailed prompts. with people.” touch-tone option. InConnect®, on the other hand, has a small vocabulary, uses 2. Reduce prompts when they occur mul- P:“Do you already have any relevant both speaker-dependent and speaker- tiple times in a single session (called qualifications?” independent recognition, permits ‘barge- ‘tapering’). C:“Not in anything to do with psychol- in’ and includes a touch-tone option that 3. Do not repeat the same prompt verba- ogy.” can be used instead of speech recogni- tim over and over (a change in prompt tion. Some traditional methods were P:“Have you thought of something to do indicates that the system is aware that used, such as scenario-based usability with caring?” the same event is recurring). evaluation, where subjects were given C:“No, but I’d like to know more about representative scenarios to follow, other 4. Provide additional information after a that. Could you tell me something, laboratory evaluations, where subjects problem (this is an important compo- please?” performed each task twice, once using nent of effective error recovery). speech and once using touch-tone, and P:“Well, it covers a number of different 5. Provide mechanisms to give the user large-scale data-collection efforts. vocations. For example, would you be control (e.g. to move from one part of interested in mental health nursing?” the interface to another, or to get help Karis found that although the two ser- when needed). C:“Quite possibly – I had never thought vices differed on various aspects, several about it before to be honest.” similar user-interface strategies were used successfully in both cases. It was 2.6 Fuzzy reasoning in com- This dialogue fragment shows different found effective to provide an initial brief puter aided telephony kinds of uncertainty. The caller is vague prompt followed, if the caller did not about his goals. Although the provider respond, by more detailed prompts. For inquirers who do not know exactly knows that about 65 % of all calls are to Effective error recovery was essential. what information they seek or have no enquire about psychology and a hint in Options include having the caller spell precise idea about which questions to this direction may be a hit, the provider is out the word, reverting to discrete vs. ask, there was until now no other option still vague about what level or kind of continuous speech input, providing more than talking to a person with a general vocation to suggest, but a specific ques- detailed instructions, or modifying the knowledge of the field. Is it possible to tion elicits the fuzzy reply “Quite pos- dialogue to improve recognition accu- design an automated computer based sibly”. racy. For example, after two failures of telephone system that can answer im- Speech Connect® to recognise the name precise questions? Up to now, computerised telephone spoken by a caller, the caller is first answering systems have been using hier- asked to choose a business type before Zajicek, Brownsey, Palau & Lippmann, archical structures, i.e. branching tree- speaking a particular company name. the School of Computing and Mathemati- structures leading from more general to cal Sciences, Oxford Brookes University, more specific questions giving a ‘hard- Karis also attempted to track callers’ UK, describe the development of a com- wired’ dialogue even though some sys- acceptance of each service as a function puter based telephone answering system tems may allow backtracking. Zajicek et of speech recognition performance, and designed to answer initial calls to the uni- al. wanted the dialogue to retain some to determine how low the speech recog- versity concerning courses on offer for key features of the human-to-human dia- nition can be and still result in a success- associate students. The central problem logue. A good information provider ful interaction and positive evaluation of was uncertainty. The system conducts a should be discursive with an inquirer the service. Karis found that the accept- question and answer type dialogue to who has unclear goals. Suggestions able recognition level depends strongly establish which areas the callers might be should reflect what has happened in the on the efficiency and success rate of the interested in and give them a short list of dialogue up to that point and previous error recovery routines. Services can be suitable course module names. The experience with enquirers. successful with recognition accuracy system uses fuzzy logic [16,17] to model significantly below 90 %, provided the the uncertain reasoning in the dialogue The authors then deal with knowledge transaction has a high success rate and and the vagueness within the database representation and user modelling. interaction is smooth and efficient. In the itself [18]. Domain knowledge can be represented as InConnect® service, which provided a various elementary groupings containing touch-tone alternative, the majority of Up to now, information and a person, two ‘collective terms’ and one ‘atomic subjects in the study preferred touch- who enters into a dialogue with the infor- term’. An elementary grouping would tone, which was faster than speech. mation seeker, has provided advice. The contain something officially recognised A large minority, however, preferred uncertainty is resolved, at least in part, in such as ‘French studies’. A collective speech because they did not have to the dialogue process. An example of a term would contain more informal remember key mappings. An interesting typical dialogue between a caller (C) and expressions used by the public such as difference was found between college provider (P) is given in the human-to- ‘twentieth century culture’, ‘academic students and older adults, with college human dialogue fragment cited below: subjects’ or ‘science interests’. An students showing a strong preference for atomic term would contain names of touch-tone and older adults preferring C:“I was thinking of doing something individual course modules, such as Mod- speech recognition. with psychology.”

Telektronikk 3/4.1997 33 layers of the system. The Dialog Man- four kinds of synthetic speech based on User ager starts off the dialogue, infers what the Rec. P.85 assessment method. Each to ask about next, decides from the res- test sample, spoken in Japanese, was ponses what updating is required, assessed on five qualities: chooses what information to offer, and 1. Overall impression decides when to stop the dialogue. The Speech Speech User Model holds the latest version of 2. Acceptance Output Input the user interest model and updates it as 3. Articulation necessary from input from the Dialog Manager. The Information Retrieval is a 4. Pronunciation simple database query and answer 5. Voice. system. Its role is to determine whether Question Linguistic the model is sufficiently refined to offer Builder Interface He found that the ∆θ-parameter had the advice or if the end of the dialogue has greatest influence on the opinion scores been reached. The Question Builder con- for over-all impression and decided that structs questions aimed at eliciting useful this parameter was a variable of TFW- responses from the User about the terms PM and that qualities of synthetic speech Dialog introduced, by providing context to the History were expressed in equivalent ∆θ. Two Manager questions and a variety of question other experiments were performed under forms. The History unit simply keeps quiet and noisy listening conditions using track of all the (essential items) of ques- both MNR and TFW-PM references in a tions and answers. And, finally, the similar manner to the first experiment. Speech Output is simply a command line Information User His results show that synthetic speech displaying the question constructed by Retrieval Model qualities in terms of equivalent ∆θ are the Question Builder (in the future syn- more stable than equivalent Q (dB) with thetic speech is envisaged). both trained and naïve subjects.

Figure 5 The architecture of the telephone answering 2.7 Assessing quality of system showing the various components arranged around synthetic speech 2.8 Voice reaction time in the central dialogue manager (after [18]) interactive services The International Telecommunication Transition from automatic interactive Union (ITU) has investigated subjective telephone services where users respond methods for assessing synthetic speech by key presses to interfaces where users quality [19,20,21]. According to ITU-T respond by speech raises the question if ern French Literature, Quantum Physics, Recommendation P.85, the references for the psychological mechanisms involved Business Ethics or Introduction to subjective testing of speech quality in- in the two ways choosing and responding Spreadsheets. For knowledge representa- clude natural speech, modulated noise are comparable. When making choices, tion the authors used a simple fuzzy reference (MNR) and time-frequency these can be signalled by speech or press- database consisting of a single relation R. warped (TFW) speech. Although such ing keys. In a classical experiment Hick The user interest model consisted of a set references are important for obtaining [23] found that reaction time increases of pairs created by assigning a user inter- stable assessments, Rec. P.85 does not logarithmically with the number of est value from a linguistic scale for each clearly state which reference should be stimulus-response choice alternatives; term in the knowledge domain. These used or how TFW speech should be gen- universally known as Hick’s Law. values of the scale were: erated. • Not interested With interactive telephone services, Toshiro Watanabe, NTT, Japan, proposes choices are presented audibly rather than • Not very interested a set of speech samples generated by visually. This corresponds to a situation time frequency warping based on phase • Undecided described by Davies et al. [24] who modulation (TFW-PM) from low bit-rate experimented with auditory stimuli and • Quite interested sampled speech [22]. He suggests that voice responses (described as ‘com- speech quality depends on three variable • Interested. patible’), while responses in traditional parameters: interactive telephone services are manual The user interest values would be up- 1. Number of PARCOR synthesis co- (i.e. ‘incompatible’). dated as a result of the dialogue. Initially, efficients (p) all values were set to ‘Undecided’. Although responding vocally to auditory 2. Frame length (L) stimuli may appear ‘natural’, it does not The system architecture of the eight com- 3. Maximum phase shift (∆θ). necessarily mean that it is ‘compatible’ ponents is shown in Figure 5. The User (as defined by Davies et al.). Since the accesses the system by typing in ques- In order to determine one parameter vari- ear and vocal systems utilise different tions to the Speech Input (in the future able and optimise the other parameter parts of the central neural system, there speech recognition is envisaged). In the values, he performed a subjective experi- are no direct neural links between them. Linguistic Interface the input is trans- ment using TFW-PM references gene- Therefore, any strong relationship that formed into objects usable by the deeper rated by combining those parameters and may exist is the result of over-learning.

34 Telektronikk 3/4.1997 Opposite to motor responses, comparably Cristina Delogu and Andrea Paoloni involving linguistic knowledge. This little is known about vocal responses. from Fondazione Ugo Bordoni in Italy could have cancelled any effects of the Thomas [25] from Philips Corporate have addressed the task of evaluating unlike speech qualities of the two voices. Design, Netherlands wished to determine speech synthesis systems for telecommu- To test this hypothesis the authors per- whether vocal reaction times follow simi- nication services [28]. formed an experiment where the reaction lar patterns as motor reaction times. He times to recognise computer generated replicated an experiment by Mowbray They find that the nature of the speech clicks while listening to texts read by nat- [26] where subjects had to name numer- material, as well as the specific tasks ural and TTS-generated voices was mea- als from a familiar response set (their given, both have implications for sured. The results showed that reaction native language) and from an unfamiliar evaluating synthetic speech. The authors times for clicks embedded in synthetic set (a strange language). His hypothesis review some traditional tasks used for speech were longer than in natural was that vocal choice reaction times for evaluating speech quality: speech, while no difference in compre- the familiar response set (in casu English hension was found. This suggests that • Transcription (writing down what is numerals) would remain constant over synthetic speech leads to a greater work- heard) number of choices, while the choice re- load on the listener. action times for the unfamiliar set (He- • Preference judgement (giving relative brew numerals) would show a logarith- preference between alternatives) In an attention test, fluctuations in vigi- mic increase consistent with Hick’s Law. lance (viz. to indicate 150 clicks) were • Scale judgement (indicating preference measured for 22.5 minutes long texts on Lickert-scales, e.g. 1 – 7) The results (see Figure 6) clearly support read by natural and synthetic voices. The this hypothesis. With unfamiliar • Comprehension (answering multiple- subjects also had to answer questions (Hebrew) material the mean response choice questions about material). about the texts. The results showed that times grew logarithmically over sets of subjects listening to a human voice two, four or eight digit choices, confirm- The authors claim that experimental reveal cyclic fluctuations of attention of ing Hick’s Law, whereas with familiar findings from laboratory tests cannot about 3.5 minutes between ‘power (English) material the mean response make reliable predictions about TTS syn- peaks’, while those listening to a syn- time for two, four and eight choices thesis performance in real life applica- thetic voice had cycles of about 2 min- remains constant, which is consistent tions since these do not take into account utes. Thus, subjects show a more rapid with Mowbray’s finding that no appre- significant cognitive factors involved in fluctuation of attention when listening to ciable differences exist when the stimu- the perception of the synthetic voice. lus-response material is over-learned. They also speculate about merits of alternative new methods, such as measuring Thomas deduces from these findings that the cognitive load in perceiving a mes- responding vocally, rather than manually, sage, measuring reaction time, i.e. the should have similar information process- time to write down the message in tran- ing requirements. This conclusion, how- scription tasks, and concurrent tasks, i.e. 200 ever, remains to be tested. It is question- performing a second task while listening able whether just replacing key presses to the message. Hebrew Hebrew with vocalisations will be sufficient to 160 achieve an acceptable speech interface. The authors present the results of several He maintains, however, that with in- laboratory experiments where listening creased familiarity with the response difficulty, comprehension, cognitive load 120 material and as speech technology and listening attention to synthetic matures, indicating choices by speech speech (Italian) was measured. These will become as intuitive as pressing keys. show that target words embedded in sen- 80 tences spoken by a clear natural voice English 2.9 Evaluating synthetic and by the best vocoders were recognised faster than words in sentences uttered by speech 40 format-based TTS systems. However, when re-tested a week later the voices (1/100 sec) Time Mean Reaction Today few interactive touch-tone tele- from the TTS systems showed a larger phone services employ synthetic speech, 0 decrease in reaction times than the natu- i.e. voice produced by a TTS (= Text-to- ral voice and the good vocoder, suggest- choice choice choice speech) system where input of a string of ing that performance in voice recognition between between between text characters is converted into output of will increase with familiarity with syn- two four eight speech waveforms. This is mainly due to thetic speech. digits digits digits the fact that synthetic speech is still of much lower quality than recorded natural Speech comprehension tests with a Figure 6 Mean choice reaction times in 1/100 sec. speech. The main problems are [27]: natural voice and a TTS synthetic voice, (ordinate) plotted as function of number of digits to • Intelligibility with multiple choice question responses, choose between (2, 4 or 8, abscissa) with responses in • Comprehension did not show any significant differences English (I) and Hebrew (N) as curve parameters. • Listening fatigue in the answers between the two condi- Hebrew responses, which are still in the process of being • Increased attention tions. The authors speculate that the sub- learned, conform to Hick’s Law, while English responses, • Cognitive load. jects may have used different strategies, which are well known, do not (after [25])

Telektronikk 3/4.1997 35 a synthetic voice than when listening to a linguistic, geographical and historic con- 3 Video communication natural voice, but exhibit no differences text, which may account for differences in comprehension of the text passages. in intelligibility. No differences were and multimedia found in task completion between the Interpersonal video communication, e.g. Delogu and Paoloni also describe an two groups. In general, the users’ accept- video conferencing, video telephony and evaluation of an Italian automated ance of the TTS service was low and now video on the PC and via the Internet, ‘reverse’ telephone directory assistance overall impression was poor: the subjects has become possible with the availability service (ARDA). The user enters a tele- reported that the sound was not clear, of wider bandwidths and efficient picture phone number via the keypad and the pronunciation was annoying and the coding algorithms, and is now being TTS system retrieves the listing informa- voice was unpleasant. taken up in an in creasing number of tion from a database and presents it over applications. This has raised a host of the telephone. Greetings and information In conclusion, Delogu and Paoloni new human factors issues, some of them about the service are given by pre-re- observe that there were at least three are dealt with in a companion article in corded natural speech, while surnames, important differences between the labo- this issue of Telektronikk (Human Fac- proper names, street addresses and names ratory conditions and the field test condi- tors in Videotelephony). Other human of cities are presented by TTS synthetic tions which may all have large influence factors issues dealt with here include speech. This service was compared to a on the results: ambient noise (the labora- such areas as, trust in talking heads, simulated service using a natural voice. tory was at least 20 dB quieter than an evaluation of subjective picture quality, The subjects were to write the surnames office and 40 dB quieter than a kiosk in use of video-communication for deaf and addresses associated with 20 dialled the street); prosody (surnames need cor- people using sign language, avoiding the numbers. rect pronunciation to be recognised) and ‘puppet theatre’ effect in video displays, task complexity (in the laboratory one 3D pointing devices in stereoscopic dis- Their results show that natural speech name at a time was presented, while in plays and sign language interpreting ser- gave 20 % more correct surnames and the field the next piece of information vice on broadcast digital TV. addresses than synthetic speech. This is was given by the system while the sub- not very surprising since surnames, and ject was still writing the previous infor- to some extent street and place names, do mation, thus creating interference). They 3.1 User trust in ‘Talking Heads’ not always follow orthographic rules. end by asserting that more research Also, there are no rules for where to put should be directed towards the under- Walker, Sproull & Subramani [29] iden- the stresses when pronouncing surnames lying cognitive processes that take place tified three important issues about the use and here synthetic speech systems are when listening to speech to interpret the of ‘simulated personae’ (talking heads) more at a disadvantage than a human evaluation results of TTS systems and in human-computer interaction: 1. Are speaker is. Surnames tend to be isolated make informed decisions on future appli- people actually willing to interact with a words, while addresses have stronger cations of synthetic speech. talking head; 2. Will people be so dis- tracted by a talking head that it will interfere with task performance; 3. How ‘human’ is a talking head and will it The Maze Game evoke any ‘social response’. To address these questions, Taylor, McInnes, Foster & Jack [30] from Centre for Computation Interface Research, Uni- versity of Edinburgh, performed an experiment to assess user trust in talking heads as information providers. The experiment scenario was a video screen- based computer game where the players were to walk through an underground maze of rooms and passages and collect Level 1 Score=4 silver coins (value, 1 point) and gold coins (value, 100 points). Three different advisory agents appear, each in a separate game, and are compared for user trust (a control game with no advisory agent was also included). At each turn in the game the players found themselves facing two passages, one to the left and one to the right (see Figure 7). To collect the silver coins hidden in one of the two left right passages the players were prompted at CHOOSE NOW the bottom of the screen to choose ‘left’ or ‘right’. On some occasions a gold coin Figure 7 An imitation of the Maze Game screen display showing a talking head and a prompt at the was hidden and a speaking ‘agent’ ad- bottom of the screen to choose the left or right side (after [30])

36 Telektronikk 3/4.1997 vised players to choose ‘left’ or ‘right’ • Talking head, 48 % under audio-visual discrepancy (audio (except for the no agent control condi- • Disembodied voice, 46 % and video out of sync.). tion). Three different ‘agents’ were used: • Real video head, 45 % • Control condition, 37 %. Contin*, Quacchia*, Galati#, Miceli# & • Talking head agent (synthetic video) Turi# [33] from *Telecom Italia, CSELT, • Disembodied voice agent (voice only) The control condition, thus, gave the Turin, Italy, and #University of Torino, least number of departures, while the dif- Italy, have pursued this basic idea in a • Real video head agent (video of real ferences between the three agents are series of experiments to find a method person). small. The difference between the control for subjective evaluation of audio-visual condition and the mean of the three agent quality. The test material consisted of six At the start of the game, players always conditions was significant (p = 0.041) as video excerpts with audio, each lasting found the silver coins by choosing con- were the differences between the control for 15 seconds. These excerpts were sistently the same side for each new condition and each of the three agent compressed by a commercial MPEG-2 move. When the agent appeared, it gave conditions. codec and recorded in various combina- wrong advice on where to find the gold tions source quality and of coded audio coin (always the opposite side of the The time to make the decisions were and video bit-rates (256 kbit/s, 128 kbit/s silver coins), i.e. if the player followed measured and averaged, and the ratio of and 64 kbit/s for audio, and 12 Mbit/s, the advice no gold coin was found, but the average decision time for the first 8 Mbit/s and 4 Mbit/s for video). neither was a gold coin found if the gold coin and the mean decision time for player ignored the advice. After some the silver coins was computed for each Pairs of sequences of the same material more silver coin moves where players condition. The control condition yielded were presented according to the ITU-R always found coins on the same side as significantly shorter decision time ratios Double Stimulus Continuous Quality before, the agent appeared once more to than any of the three agents. Scale (DSCQS) protocol. One sequence give advice on which side to find the in each pair was always uncompressed gold coin. Taylor et al. conclude that advice does (source). The subjects were requested to make a difference: there were more evaluate each sequence on three quality For players who had followed the agent’s departures from immediate prior experi- parameters; 1) ‘colour reproduction’, advice on its prior appearance the advice ences in the advice conditions than in the 2) ‘smoothness of movement,’ and was now pre-arranged to be correct. If control condition, and the players who 3) ‘picture sharpness’, using the con- players again followed the agent’s advice were advised took longer to reach their tinuous ITU-R quality scale for the (in the face of contrary experience of decisions. DSCQS method. The reproduction of misplaced trust) they were rewarded with ‘low’ and ‘high’ audio frequencies was a gold coin. Players who did not follow With respect to the decision on the first also assessed. In addition to the audio the advice did not receive any coin. gold coin in each game, there were mod- and video evaluations, an evaluation of erately significant differences between the ‘overall audio-visual quality’ was For players who had not followed the the talking head on one side and the dis- requested. agent’s advice on the first occasion, the embodied voice and real video head on advice was now pre-arranged to be the other side. It appears that the talking The results of Contin et al. [33] show wrong. If they followed the advice they head was more persuasive than either the that for any video bit-rate, high fre- were not rewarded, but if they again real video head or the disembodied voice. quency audio quality increases with rejected the agent’s advice they received The authors speculate that the differences increasing audio bit-rates, and for any a gold coin. The game was thus designed found between the talking head and the audio bit-rate, the subjective ‘picture to ensure that each player was given cor- other conditions may suggest that a suffi- sharpness’ increases with increasing rect advice once and wrong advice once ciently realistic human-like synthetic video bit-rates. This suggests that both to counter any bias either against or in ‘talking head’ might be trusted more. audio and video bit-rates significantly favour of taking advice from an agent on influence the perceived quality of high subsequent interactions. The net result of frequency sounds. Analysis of variance the design was that players who were 3.2 Subjective evaluation of also showed that the ‘video bit-rate’ and consistent in their decision to either audio-visual quality ‘sequence’ had significant effects on the accept or reject the agent’s advice were ‘smoothness of movement’ quality, and rewarded with a total of one gold coin (in Perception of audio-visual quality of that the ‘audio bit-rate’ had a small, but addition to the silver coins collected). multimedia services, including real-time significant, effect on the quality of interpersonal communication and inter- ‘colour reproduction’. The players each completed four games active remote data retrieval services, may (one for each condition) and within each not only depend on the quality of each of For evaluation of low frequency audio game made four decisions regarding the its two principal components (audio and quality, only the ‘audio bit-rate’ and gold coin. These decisions are the main video) but also on their interactive (syn- ‘video sequence’ factors were significant. focus of the experiment. For each agent ergetic) quality. McGurk & MacDonald Surprisingly, the ‘video bit-rate’ was not type, the number of decisions that de- [31] and Sekiyama & Tohkura [32] significant for the ‘overall audio-visual parted from the previous experience (i.e. demonstrated perceptual fusion of audi- quality’. The authors warn, however, that the side silver coins were found) was tab- tory and visual (lip-reading) information one should be careful about generalising ulated and the means computed. The per- (McGurk effect) in speech recognition that the ‘video bit-rate’ is an insignificant cent deviation for each condition were: factor for ‘audio-visual quality’. They conclude that in order to avoid any

Telektronikk 3/4.1997 37 evaluation bias, ‘overall audio-visual decreases with increasing compression Tandberg Television, was initiated to quality’ should be assessed in separate ratio up to 20 % for the lowest bit-rate. study the technical possibilities offered test sessions. The authors conclude that further studies by the standard for sign language video. are required to determine what factors The goal was to develop a permanent 3.3 Coding degradation and are important for remote interpersonal additional sign language transmission in recognition of emotion video communication at low bit-rates in parallel with the ordinary TV programme e.g. mobile networks. as an optional feature of the TV receiver. In remote interpersonal video communi- Fosse & Dolvik [40] of NRK and Tand- cation, such as video-telephony and 3.4 Sign language and 64 kbit/s berg Television, respectively, describe a videoconferencing, the visual feedback picture quality project to develop such a service. With of the gestures and facial expressions of the imminent adoption of the DVB stan- the interlocutors is the raison d’être for Several studies have shown that the dard it was deemed irrelevant to consider the medium. However, the recognition of 64 kbit/s ISDN video telephone is suffi- transmission of the service via the non-verbal information, attitudes and cient for remote interpersonal sign lan- NICAM multiplex. emotions may be lessened by the picture guage communication by deaf people degradation introduced by the video cod- (see e.g. [35]). However, it has also been The authors give a brief technical de- ing algorithms. shown that remote sign language conver- scription of the basic MPEG-2 DVB sation by video telephone is more stress- system for transmitting a user selectable Contin*, Quacchia*, Galati#, Miceli# & ful than real face-to-face conversation service where sign language is trans- Torrini# [34] from *Telecom Italia, [36,37,38]. mitted in parallel with the ordinary TV CSELT, Turin, Italy, and #University of programme and shown as a small picture Torino, Italy, carried out an experiment Kamata, Yamashita & Hirama [39] of in the upper left hand corner of the TV to identify the thresholds for correct Utsunomya University, Japan, have screen. Thus, hearing people will not identification of human emotions at very studied the relationship between picture have their viewing disturbed by the sign low bit-rate coded video sequences. quality of 64 kbit/s ISDN video tele- language. Twelve silent video sequences, showing phones and sign language conversation. children expressing six different emo- Using Scheffe’s paired comparison Two prototype satellite receivers were tions (but not showing any context), were method they performed perception tests developed to demonstrate the system. compressed by the MPEG-4 video com- with deaf and hearing subjects and differ- The system has been demonstrated on pression algorithm and up-sampled to the ent picture quality. several occasions. Its general configura- Common Intermediate Format (CIF) tion is shown in Figure 8. The pilot pro- (360 pixels, 288 lines, 25 Hz, 4:2:2). The results show that the deaf subject ject has proven the practicality of the Three bit-rates, 16 kbit/s, 24 kbit/s, preferred pictures with higher temporal proposed system. Work will now concen- 48 kbit/s, and the uncompressed CIF resolution (faster frame rates) to pictures trate on implementing the system and sequences were used for each video with higher spatial resolution (more find industrial partners who will put the sequence. Subjects had to indicate which detail), while the hearing subjects pre- necessary consumer equipment on the of the following emotions the child in ferred higher spatial resolution to higher market. each sequence expressed: temporal resolution. These tests did not actually use real sign language conversa- 1. Anger tion, but a simulation, and only one deaf 3.6 ‘Informal’ video-communi- 2. Joy subject, but the authors are confident that cation 3. Disgust their findings are representative for deaf 4. Surprise people using sign language. The authors Informal communication designates those 5. Sadness suggest that the party receiving the video spontaneous, non-mediated, informal 6. Fear picture should be able to control the face-to-face exchanges, usually between 7. Other. video coding parameters of the sending two or three people, that occur in the party to suit their special requirements. work place in chance meetings around Subjects also had to rate the intensity of the office, during coffee breaks, in the the emotion shown and the degradation tea-room, or in informal gatherings at of each video sequence on two Lickert 3.5 Sign language in digital conferences, and that fulfils important scales. video broadcasting functions such as information exchange, trying out ideas, harmonisation of atti- In general, the results show that lower Real-time sign language interpretation in tudes, etc. Most non-mediated, informal compression bit-rates lead to lower a small window in the broadcast picture communication is characterised by being: correct recognition of emotion, but not to of TV programmes has been provided for the same extent for all emotions. And the deaf in Norway on special occasions, • Unscheduled even in the non-compressed versions of e.g. political debates in connection with • Confidential the sequences used, correct recognition general elections. The Norwegian Broad- • In real time depends on the emotion (ranging from casting Corporation (NRK) has for sev- • Interactive 63 % for ‘surprise’ to 92 % for ‘joy’). eral years investigated the possibility of • No agenda Perception of the intensity of the emo- offering sign language video as a perma- • No moderator. tion, however, does not seem to be nent service. With the advent of the affected by the compression. As ex- digital video broadcasting (DVB) stan- To derive user requirements for telecom- pected, the subjective picture quality dard, a co-operation between NRK and munication systems that support informal

38 Telektronikk 3/4.1997 communication, Suwita and Mühlbach imagine informal communication by helps to solve every-day work problems [41] from Heinrich-Hertz-Institut für means of video communication. in an efficient manner. Negative effects Nachrichtentechnik, Berlin, studied the were also mentioned; e.g. that people nature of informal communication by The study confirmed the authors’ belief may be disturbed in their work and that it interviewing ten male and nine female that informal communication was con- may upset the social climate by gossip employees from different work areas sidered important, even indispensable for and rumours. such as, court of justice, school, research work. Besides improving the social work institution, hospital, industry, etc. In the climate and the subjective well-being of Among the most crucial conditions for semi-structured interview, some ques- the employees in general, various posi- informal communication to take place are tions centred about occasions, locations, tive aspects of informal communication confidentiality and privacy; people some- durations, benefits and disadvantages of were referred to. These include bringing times make statements and express informal communication. There was one team members up-to-date, that it stimu- opinions during informal chats that they question whether the interviewee could lates new ideas and thoughts and that it would rather not make in a formal meet-

Additional sign language service

MPEG2 CODER MPEG2 transport stream

MULTIPLEXER MODULATOR

TANDBERG TT1100 DEMOD and DEMUX

MPEG2 CODER

Ordinary program material - live or taped

PAL monitor with sign PAL monitor sign language picture language is chosen in upper left corner not chosen

Figure 8 The Digital Video Broadcasting (DVB) program chain showing the addition of a real-time sign language interpretation service signal to the main programme material, and the viewer’s option to select or reject the live sign language interpretation service provided in the upper left corner of the TV screen (after [40])

Telektronikk 3/4.1997 39 ing. Environmental factors that support cation systems for tele-co-operation and present scenes at some distance ‘behind’ informal communication are spatial distance work should take this into con- the screen (virtual image) to avoid the proximity, availability of common rooms sideration. The authors also expect that miniaturisation effect in an autostereo- and open office doors. Informal en- systems that support informal communi- scopic (3D) desktop display for commu- counters last only a few minutes, seldom cation will enhance user satisfaction and nication applications. They report on a longer than 15–20 minutes, but may improve work efficiency. study where they shifted the plane of recur during the working day. accommodation with a collimating 3.7 The ‘puppet-theatre’ effect optical system involving Fresnel lenses On the question of video-communication, in 3D desktop displays and spherical mirrors (see Figure 10). the majority of interviewees could hardly imagine to have informal confidential Prussog et al. [43] had found that partici- In the common 3D desktop displays, only chats by means of video-communication. pants in stereoscopic videoconferences a limited field of depth around the screen They felt that it did not provide enough felt uncomfortable wearing 3D goggles surface can be utilised without causing social presence and expressed grave con- or other head-mounted equipment. The eyestrain. This leads to people and faces cerns that video-communication does not authors therefore conclude that 3D dis- being reproduced at an unnaturally small offer sufficient privacy and confi- plays should be autostereoscopic, i.e. scale, which causes a ‘puppet-theatre’ dentiality, e.g. fear of being observed that the image can be viewed in 3D with- effect (see Figure 9), preventing a high (even videotaped) or having no control of out any supplementary devices like degree of presence in interpersonal com- who could watch them. shutters, polarising glasses or similar munication applications. gadgets, while allowing the viewer some The authors conclude that informal com- freedom of movement. Their solution is Hopf & Runde [42] of the Heinrich- munication is very important in the work to place an optical collimating system Hertz-Institut in Berlin describe a way to context, thus designers of telecommuni-

virtual image

"puppet theatre" effect

optics

Figure 9 The principle for the adaptation of acommodation distance by means of optics to defeat the ‘puppet theatre effect’ (a = accommodation distance, c = convergence distance). The small image on a screen can be experienced as a life size ‘virtual image’ if the accommodation distance of the visual system can be tricked into perceiving the image as being further away than it actually is (after [42])

40 Telektronikk 3/4.1997 spherical mirror

collimation optics

video camera

beam splitter polarizing filter

beam splitter

LC-panel LC-panel mirror lens

lens

light source before the 3D-display unit, and they eval- the mirror uate three different optical realisations. triplet display The simplest solution is to place a con- size) thus unit vex lens in front of the display. This will causing mini- beam splitter make the virtual image appear to be mal chromatic larger and further back than the screen aberration, and a shift surface. It is important to find the right of eye-position of 150 mm produces an Figure 10 A 3D desktop display unit for parameters. enlargement of only 1.8 of the pixel size. multimedia. The collimation optics Some curvature of field makes it neces- (upper part) will project a virtual 3D To avoid eyestrain, the accommodation sary to change accommodation for view- image from the display unit (lower part) plane should be placed some distance ing different parts of the screen; the vari- so as to increase the perceived accom- behind the screen so that differences be- ation in focus amounts to ±150 mm. modation distance and defeat the ‘puppet tween accommodation and convergence However, the beam splitter in the on-axis theatre effect’ (after [42]) are minimised. This should result in an spherical mirror system causes a loss of increase in the visible volume of depth, about 75 % of the light. Geometric dis- thus escaping the ‘puppet theatre’ effect. tortions arise in all three systems, but the The other optical solutions evaluated best results were obtained with the on- were; a concave spherical mirror and axis concave spherical mirror; observing beam splitter, and an off axis (45°) con- from an on-axis position leads to distor- cave spherical mirror. tion of only about 0.6 %. three co-ordinates in 3D space to create a user-friendly pointing device. (Using a Hopf & Runde found that the use of a The authors conclude that the on-axis mouse has the added advantage that ordi- Fresnel lens gave unacceptably large concave spherical mirror and beam nary 2D computer applications can be chromatic aberrations even at low magni- splitter is the best system. It produces no performed without changing pointing fications and on-axis eye-position (the chromatic aberration or distortion and device.) difference in displacement of the red and resolution in on-axis viewing is accept- blue components of white light was six able; the loss of light, however, must be Runde & Suwita [44] from the Heinrich- times the pixel (triplet) size). The off- compensated for by increasing the dis- Hertz-Institut, Berlin, have devised and axis spherical mirror system reduced the play luminance. evaluated three different 2D-to-3D point- spatial resolution, even at low magnifica- ing devices for use with interactive tions, by enlarging the pixel (triplet) size 3.8 2D-pointing in 3D space stereoscopic desktop displays. These are over four times. With off-axis systems or called the ‘Surface pointer’, the ‘Floating high magnifications, deviation from cor- pointer’ and the ‘Torch’. In stereoscopic Addressing objects in 3D space in inter- rect focus leads to asymmetrical enlarge- images the pointer (3D cursor) has six active co-operative stereoscopic applica- ment of the pixels (astigmatism). degrees of freedom: three translations of tions requires stereoscopic pointing. The movement in the three orthogonal dimen- most used pointing device is the standard The use of an on-axis concave spherical sions and three rotational directions for computer mouse. But how are the 2D co- mirror and a beam splitter gave the changing orientation. The mouse, how- ordinates of the mouse translated into smallest pixel magnification (only ± of ever, has only two degrees of freedom,

Telektronikk 3/4.1997 41 movement along the two co-ordinates of on the surface of the nearest object. Left- 4 Design for all – support the plane, plus one, two or three mouse right and towards-away mouse move- buttons for clicking on various files and ments are translated into rotation of the for elderly and disabled applications. torch around the vertical and horizontal An area that is now receiving increasing axes, respectively. The position of the attention is the accessibility to infor- The ‘Surface pointer’ is a three-dimen- torch in 3D space is altered by moving matics and telecommunication systems sional arrow that can be moved by the the mouse while pressing a mouse (ITC) by elderly and disabled people. mouse in the two dimensions of the button; (left button: positions in a plane The Commission of the European Com- screen plane. The third (depth) dimen- parallel to the screen surface, right munities (CEC) has mandated a number sion is automatically adapted so that the button: positions in depth). This pointing of research projects and standardisation pointer always appears on the surface of system relies on detailed information activities to rectify this situation. It is not the objects it touches. If the pointer is about the surfaces of the objects in the acceptable in a modern society that up to moved away from an object, the pointer picture. one quarter of the population cannot use does not jump to the background, but common telecommunication and infor- keeps to the same plane as the object. Since two of the pointers need informa- mation services. The pointer can be rotated in 30° steps in tion about the surface of the objects their two planes by pressing the left and right use is restricted to computer-generated With increasing standards of living and mouse buttons, respectively. This point- images like Virtual Reality scenes. improving medical treatment, people live ing system relies on information about Therefore, all three systems were longer and are in better health than ever the 3D surfaces of displayed objects. implemented on a fast graphics com- before. With the general lowering of re- puter. A list of pointing requirements and tirement age coupled with larger pen- The ‘Floating pointer’ is of a three- hypotheses with regard to the three sions, people are better off economically dimensional arrow. Movement of the pointers was set up for testing: and have longer active retirement time mouse determines the position of the 3D than was the case only a generation ago. pointer in a plane parallel to the screen The results of the tests are shown in These senior citizens make up the most surface. Moving the mouse towards or Table 2. They testify that, generally important new customer groups (e.g. the away from the user while pressing the speaking, the subjects positively eval- fastest growing crowd of Internet users in left mouse button changes the plane of uated all three 3D pointing systems. One the USA are retired senior citizens who depth in the 3D picture. Moving the obvious result is that for non-moving have more time and money to spend than mouse while pressing the right mouse objects the ‘Floating pointer’ was eval- many other groups). button will change the orientation of the uated as better than expected. The pointer. This pointing system does not authors cannot recommend any one There are also large groups of people rely on information about the 3D scene pointing system as such, but that the with various impairments, which are not and is therefore easy to implement as an choice of pointing system is made as visible as the ‘classical’ disabilities ‘add-on’ to current 3D systems, in- according to the specific task or applica- such as deafness, blindness or motor cluding live stereoscopic video. tion. However, the authors affirm their impairment. These are people with learn- main finding that it is possible to trans- ing disabilities, people with reading dis- The ‘Torch’ pointer is an image of a late 2D pointing into 3D pointing for use abilities, people who are educationally torch (flashlight to some) with a slightly with 3D PC applications and 3D video disadvantaged, people with psychiatric diverging, yellow light beam which ends applications. diagnoses and people who are culturally or socially disadvantaged. These constitute a very large number of people, who have mostly gone unrecognised in industrialised countries and who have Table 2 Hypothesis with regard to pointing requirements received little attention compared to people with the ‘classical’ disabilities, Pointing requirements Surface pointer Floating pointer Torch which often have powerful organisations and spokespersons. 1. Accurate pointing easy difficult easy The integration of disabled people into 2. Hit objects in depth easy difficult easy ‘mainstream’ society is gaining momen- 3. Point at positions in space impossible difficult impossible tum in many countries. One fundamental requirement for independent living, how- 4. Point in any direction difficult easy easy ever, is to have open access to basic 5. Point at transparent objects easy difficult impossible information and communication services on par with all other citizens. Some 6. Point through transparent surfaces impossible difficult easy people with profound impairments will 7. Move along any direction difficult difficult impossible obviously require special equipment and services, but very large groups of dis- 8. Follow any moving object easy difficult easy abled people and senior citizens will only 9. Move along any object easy difficult easy require marginal improvement of standard equipment and common services: 10. Mark any object with a tag easy difficult easy e.g. larger keys, more legible markings

42 Telektronikk 3/4.1997 and displays, improved sound quality and emerging as the universal information are displayed simultaneously so that the simpler operating procedures. It is a deep source providing information normally user can engage in any one at any time. irony that despite, or maybe because of, found in books and periodicals, or by Speech based information is essentially the many recent technological advances, attending courses and conferences, etc. sequential. The authors have devised a usability and perceived quality of most This is of great benefit to sighted people, concurrency similar to that of a graphical equipment and services have deteriorated who have instant access (well, nearly) to interface by conceptually structuring the in several respects; key-pads with re- a vast amount of information. Blind page to offer options such as lists of duced tactile feedback have become people, however, are denied this source headings, lists of links and lists of key- smaller and more ‘fickle’ to use; sound of information since most extant web words found in the text, or to read out quality has worsened (with low quality browsers are GUI-based and thus in- paragraphs of text, etc. This they achieve synthetic speech increasingly replacing accessible. Besides, it is necessary to by using a menu-bar, which is accessed natural speech); displays are often added have access to a computer for browsing by speech or function keys. without any regard for legibility (with the web, which is not common for blind low contrast, small characters, obliterat- people. Current Braille and speech screen Zajicek and Powell have carried out ing reflections, inconvenient positioning, readers are deemed quite unsatisfactory experiments to evaluate the menu-bar as etc.); operating procedures are not very [46,47,48]. The idea of a speech-based a tool for orientation on the web page intuitive or much too complicated for interface for accessing the web via the and to determine which menu-bar option, many people; operating instructions are telephone has been proposed several headings, links or keywords, was per- either absent or too difficult to read and times, but little progress has been made ceived as most effective by the users. follow for many people; the list can be up to now. From their initial experiments the authors carried on and on. believe that there is a potential for the Zajicek & Powell [49] of the Oxford presentation of web pages in audio Recent legislation concerning people Brookes University, UK, have addressed format. Although orientation and infor- with impairments (e.g. Americans with this problem. The authors declare that the mation retrieval took longer, the models Disabilities Act of 1990 and Telecommu- cognitive demands of browsing the web created by the users from the available nications Act, now going through Con- with a graphical user interface are con- information seemed comparable with the gress) has put these issues in focus, siderable and require total attention. models achieved by visual scanning of a especially in countries like USA that They envisage a future in which the web page. Users can become impatient if have the most liberal markets, and it is requirements of continual update of they have to listen to output they are not now only a matter of time before we will information (e.g. when travelling, when interested in. By providing the menu-bar see similar legislation passed in Europe shopping, or while performing leisure for navigation the users can jump around (either as national laws or issued as activities) increases the need for indi- on the web page. The authors are con- Directives by the Commission of the viduals to update themselves in situations fident that if they can find a way of pro- European Communities). where they have no access to a computer. viding concurrency of information that is They thus imagine many situations where transparent to the user and embodies the However, economic concerns may soon it can be more useful to access informa- power of the concurrency inherent in dictate as much as any legislation that tion on the web via the telephone (e.g. a GUI, access to computers by telephone something has to be done since in- mobile phone) instead of being tied down will open up an important new informa- creasing market shares of elderly and dis- in front of a computer. tion channel. abled customers will be significant and cannot be ignored. The tools already To meet this challenge, the authors built a 4.2 Communication for non- exist for justifying the positive cost/ ben- helper application they called WebChat speaking users efit in improving the user interface, not that intercepts the HTML code that makes only for ordinary customers but equally up a web page and provides a conceptu- People who cannot speak are dependent so for disabled and elderly customers ally organised version of the web page on various forms of assistive technology [45]. using speech. The authors re-described such as hand-held and portable commu- the structure of a web page so that it nication devices (PDAs) for communi- could be ‘read’ out to a visually impaired 4.1 Accessing the WWW by cating. Such small devices usually have user over the telephone. This involved telephone low processing power and a link with a forming a conceptual model of the infor- more powerful computer, e.g. for word mation content of the page and its inter- The Graphical User Interface (GUI) has prediction in text production, is desirable. relationships and presenting it as speech. been a catastrophic step backward for A link to a powerful computer, e.g. over blind and severely visually impaired a telephone or mobile phone, would in A web page can be presented con- users of IT. Inherently, it is very difficult effect give the user the processing power ceptually as a collection of textual to convey all the information in a GUI of a large computer in a cheap small objects and links. WebChat provides screen in a meaningful way to a blind hand-held device. What is the minimum access to conceptually different parts of user, whereas the earlier line based sys- bandwidth requirement for using a word- the web page using function keys and tems were relatively easy to convert into prediction application before a user is different synthesised voices to indicate Braille on an 80 character Braille reading significantly delayed using a word pre- conceptually different text features, such device. diction application? as headings, links and text paragraphs. The World Wide Web (condescendingly Beattie, Hine & Arnott [50], the Micro- A graphical interface is essentially con- referred to as the World Wide Wait) is Centre, University of Dundee, Scotland, current; i.e. many items of information

Telektronikk 3/4.1997 43 Mac Quadra 700 Unix Unix Unix Workstation Workstation Workstation

PSTN Modem PPP Internet Link

Ethernet Ethernet Network Network

Figure 11 The basic experimental set-up used for all conditions in assessing the assistive process (after [50])

have performed a study to compare the more restricted variety of communicative In a preamble, the ‘Travel Agent’ first influence of different baud rates of a techniques. The conventional telephone asked which state was to be visited. The telecommunication link between a local provides a multitude of interactive ser- reply was either a name typed on the terminal and a remote computer on the vices (e.g. 24-hour banking, and other ‘text telephone’ or the display of a state usability of a text prediction application, services) which do not require speech map on the terminal screen. This was and to compare these to the use of text input. Many non-speaking people regu- followed by a request for the first town to prediction as a local application. larly use text telephones and fax services, be visited. The reply was either a typed but studio or desktop video-telephones name or the positioning of a box marker Rather predictably, they found that the are also being used as an alternative to on the map to indicate the desired town. interval between the key press to re- face-to-face meetings. The Internet and The ‘Travel Agent’ requested confirma- ceiving the prediction rose as the speed multimedia services hold high promise tion by repeating the chosen name in a of the connection fell (see Figure 11). for non-speaking people. text window, and the subject confirmed The delay at the two slowest baud rates by either typing “yes” or placing an (1248 ms at 1200 baud and 800 ms at The basic problem, however, is that text arrow in the box on the map. If the 2400 baud) of the remote applications input instead of speech is very slow. In ‘Travel Agent’ responded with an in- were unacceptably long when entering conversation and communication pictures correct name, the subject either retyped text and waiting for a prediction to be and graphics can be used to convey the correct name or again placed the box displayed. The authors conclude that complex ideas and various symbol and marker around the desired town on the word prediction over a remote link can picture systems have been devised for map. be used effectively if the transmission Augmentative and Alternative Communi- rate is 9600 baud or higher. Below this cation (AAC). Five experimental episodes where the rate the user notices the delay which ‘Travel Agent’ asked for the next town to leads to frustration and errors. The large To test the practicality of graphic com- be visited and the road to get there fol- processing power required by some munication as an alternative to speech, lowed this preamble. In each case the assistive techniques may thus be avail- Hine, Beattie & Arnott [51] of the Micro- subject either responded by typing the able to users by a cheap, small portable Centre, University of Dundee, Scotland, names via the ‘text telephone’ or by indi- device linked to a remote server, pro- compared the time taken to exchange cating on the screen map. The time to vided the link has sufficient speed to sat- information in text form and in picture complete each episode was recorded. The isfy the use of the application. form. Their hypothesis was that some full experiment consisted of eight sub- users could exchange information at a jects doing four sessions each with two 4.3 Picture annotation for non- significantly faster rate by displaying and text and two map conversations. All sub- speaking people annotating pictures than by typing the jects had the same eight scenarios and the same information. text and map sessions were counter- balanced to avoid any order effects. As telecommunications have become The experiment involved a scenario more and more important, people with where eight subjects indicated to a The results are very variable and can speech impairments are even more dis- remote ‘Travel Agent’ a number of only hint at any tendency. Five of the advantaged than before. In a face-to-face places they wanted to visit in their holi- eight subjects completed the text tasks situation, a non-speaking person can days, using either a text telephone (simu- faster on average than the map tasks. always resort to writing, pointing, pan- lated on a computer) or by displaying and However, the eight subjects varied with tomime or pictures. In telecommunica- annotating a map on the computer screen. respect to speech and upper limb motor tions, a non-speaking person has a much

44 Telektronikk 3/4.1997 impairments, and used various screen being introduced with no regard for the tests were held in the students’ pointing devices (mouse, roller-ball, chin people with impairments. The quality of rooms, otherwise a suitable location was controlled roller-ball, head-switch driven work in a study is no guarantee for its found. In the first test the subjects were mouse and keyboard, on screen key- impact on the practices of the operators asked to set up the mobile equipment, board, concept keyboard and Keyguard) and the markets, the findings and results write an e-mail off-line and send it using making any comparisons and conclusions are duly acknowledged, but nothing hap- Pegasus e-mail application. In the second invalid. Although the authors explain pens to alleviate the situation for the dis- test the subjects had to search for some some of the variability and find consola- abled user. specific information on the Web using tion in the fact that at least three subjects Netscape® navigator. The time to com- did better with pictures than with text, The authors conclude that the process plete the various tasks was logged by the supporting their original hypothesis that outlined in their report allows user re- computer and the tests were video filmed “some users can exchange information quirements for the usability of telecom- with time registration. …”, the results must actually be regarded munications to be analysed, problems to as highly inconclusive due to flawed be highlighted and solutions to be The time needed to connect the GSM method and bad experimental design. It suggested and investigated so that issues handset and install the PCMCIA card in should be interesting to see results for can be flagged for a variety of users the laptop varied from 40–60 seconds for text vs. picture communication that are under consideration. In general, this subjects with no upper limb impairments truly comparable, with the right screen method illustrates the wide range of to 80–100 seconds for subjects with more design it may well be that communi- factors that may influence the usability of severe upper limb impairments. Most cating by pictures rather than by text is a a telecommunications service, particu- subjects reported that a built-in PCMCIA better proposition for many users and larly for users with disabilities. card would significantly simplify set-up applications. of the system and make the terminal 4.5 Mobile access to WWW more accessible. 4.4 Disabilities and tele- services communication services The sending of the e-mail showed that the quality of the mobile network con- A GSM telephone connected via a PCM- nection (which varied from 0.5 to 4.0, The use of a telecommunication service CIA card to a personal (laptop) computer with a mean of 2.5) was a deciding involves interaction with a complex set moving about in a cellular radio network factor. The second test required an even of elements and performing a variety of is a technology that will enable ‘mobile’ higher network quality. All subjects com- activities. Modern and future telecommu- access to the Internet. This may provide a plained about the noise due to the inter- nication services will allow using a number of novel opportunities, such as ference between the GSM handset and variety of media, both in the control of library consultation, distance education the speaker output of the notebook dur- the services and for the information and banking services, to people who are ing data communication. The audio out- exchange. If the user has a disability the mobility or motor impaired. These bene- put and PCMCIA slots were only 10 cm service may be unusable. The barrier fits, however, can only be realised if the apart and the GSM handset cable was too may lie at three levels; in the terminal, in equipment is fully accessible to the short to have the GSM phone at suffi- the network or in the service (see [52]). intended users, i.e. any user barriers in cient distance from the audio output. the terminal, network and service must More serious, though, was that normal Hine, Beattie & Arnott [53], the Micro- be removed as early as possible in the control of the cursor was lost if the Centre, University of Dundee, Scotland, design stage. mouse or mouse emulator was on the present an interim attempt to develop a same side as the GSM handset, especially method of anticipating and recording the Van Mele, Edwards, Verheust & during long connections with the Web problems that users with disabilities face Spaepen [54], Katholieke Universiteit, server. The handset had to be moved to when using telecommunication services Leuven, Belgium, wished to determine the opposite side and at least 40 cm away and the potential solutions that could the feasibility of mobile access to the from the laptop before normal control of address their problems. Their analysis is WWW by mobility impaired people using the cursor could be resumed. Also, inter- part of a wider issue of finding services a laptop computer connected to a GSM ference with the GSM handset affected that users need and the need for adapting handset. The authors believe that mobile the operation of a hearing aid and arm current and future services to the users’ telephones will serve an important role as prosthesis. needs. Their analyses and procedures for assistive technology for mobility im- assessing user requirements and pro- paired people, i.e. people who are bound The task of writing and sending an e-mail posing usability improvements of present to a wheelchair, who have difficulties using Pegasus was completed in between and future telecommunication services leaving their home or who cannot easily 181 and 685 seconds. The task of search- derive from the ACTS UMPTIDUMPTI access a library or come to classes. ing for information on the Internet using project and follow accepted mainstream Netscape was completed in between 122 methodology for usability assessment. The authors used a system consisting of a and 673 seconds, while the number of Similar studies and analyses have been Nokia 2110 I GSM mobile telephone and selections necessary to find the required carried out both in past and present pro- PCMCIA card connected to a T400 CDT information varied from 13 to 100. The jects (e.g. TUDOR, COST 219, TIDE, notebook laptop computer. Ten students high values are mainly attributed to one Interact, Include, ETSI PT6V, etc.) with- (aged 19 to 32) with various motor subject who in addition suffered from out much impact on service operators, impairments served as subjects. All sub- tunnel vision and thus had great diffi- manufacturers or regulatory authorities; jects had previous experience with com- culty reading the screen. Post-test ques- new services and equipment are still puters. Where there was GSM coverage, tions revealed that 90 % of the subjects

Telektronikk 3/4.1997 45 favoured using these applications profes- The videotape contains nine video text telephony relay services. Although sionally. sequences designated: the results reported are somewhat inconsistent and fragmentary, a general picture 1. Silent voice (reference video with The authors conclude that the trials have emerges showing that deaf signing American Sign Language) demonstrated the feasibility of mobile people may benefit from all the systems GSM/laptop set-ups for people with 2. Fast (rapidly signed language) tested, and that the usefulness of any motor disabilities. However, the tests video system improves with increasing 3. Slow (slowly signed language) have also revealed significant user bandwidth, i.e. higher picture frame rates barriers on all three levels; terminal, net- 4. Frame rate (picture update rate, and/or higher spatial resolution. The work and service. It will be necessary to >12 frames/s, 20 f/s is preferable) issue of deaf people, either pre-lingual reduce interference and simplify opera- deaf sign language users, or adventi- 5. Static resolution (striped still patterns tion of terminal equipment, to improve tiously deaf people who can speak, read for resolution measurements, QCIF GSM coverage and to speed up access to and write, using telecommunications has (176x144) acceptable, CIF (352x288) services to make mobile access practical been subjected to many research projects, good) for motor impaired users. which all tend to gravitate towards the 6. Blur (sequence for judging blur of same results as presented above. 4.6 Video quality for sign details in sign language) language communication 7. Sync (sequence for measuring synchro- 5 Group working support nisation of sound and video, should Video communication offers great possi- not be higher than 0.1 s for lip-reading services bilities for deaf people using sign lan- supported by audio) Group working has emerged in recent guage. For selecting or developing video 8. Lip reading (sequence of talking per- years as a novel way of working together systems for communicating in sign son to determine usefulness for lip- on a document or other task via the language, a method is needed to deter- reading) telecommunication network. The actors mine the practicality of any system. taking part in group-work activities may 9. Delay (sequence for measuring picture be geographically proximal or dispersed, A simple method of measurement was delay (sound to be sent trough separate this is not the main issue – the main issue developed in the project “Quality re- cable) delay should be less than 0.5 s is that the system and the services must quirements on video communications for for smooth conversation and below support interactivity between several sign language” and is reported by Hell- 1.2 s to be acceptable). people dedicated to solve a common task ström [55] of Omnitor, Sweden. The via the network. Special forms of group method extracts technical factors related The validity of these quality factors has work in distributed organisations are to the requirements of sign language been verified in conversation tests by telecommuting, mobile working, work- users. The method is easy to use and proficient signing persons. Using this group computing, videoconferencing, requires only one or two video tape method, it should be possible to ascertain etc. In a companion article in this issue of recorders (VCR) and a test tape that was the minimum quality necessary in a Telektronikk (see The ergonomics of tele- produced within the project. Although prospective video telephone system for working), Bakke deals more in depth the method is based on sign language, it sign language communication. with these matters. does not require knowledge of sign language to be used. In use, the videotape In a comparison of projects using differ- is played over the video telephone sys- ent techniques, bandwidths and services, 5.1 Assessing user needs for tem to be assessed (see Figure 12). It is Delvert [56] of Kalejdo Management future work practices preferable to record the outcome at the Consultancy, Sweden, investigated the far end, but it is possible to make an merits of different systems for signing To assess the user needs of mobile pro- assessment by direct observation. deaf people, including comparison of text fessional workers, Sallnäs, Hedman & telephones and video telephones at dif- Virtanen [57] of Telia Research, Sweden, ferent frame rates and the usefulness of studied selected members of mobile pro-

TANDBERG Vision TANDBERG Vision

ISDN VCR (PSTN) VCR

Figure 12 The basic experimental set-up used for the quality measurements (after [55])

46 Telektronikk 3/4.1997 fessional teams in demanding environ- Telecommuters seemed to fall into two • All groups utilise information, both ments; a fire fighter, a building worker, a distinct categories: those who planned from internal and external sources. plumber, a salesman, and a reporter and a their telecommuting work and those who Today, information is mainly commu- production manager from a local TV see it as an opportunity to work ad hoc. nicated to mobile teams via auditory station. The topics studied were social The first group defined special telecom- channels, but there was a desire for group processes, awareness of other muting days and planned specific work more visually displayed information. members of the organisation, instant tasks for these days. They regarded their Generally, these mobile professionals team status, applicability of group soft- office days as social, devoted to meeting had only little need for processing ware, etc. people and attending meetings. The ad information. hoc group, on the other hand, avoided • There was a general desire for aware- The results of these studies show that planning and tele-worked when the need ness through shared contexts (Instant multiple methods are necessary to obtain arose to execute individual tasks. They Team Status). This is now mainly an understanding of present and antici- also often split days between office work achieved through auditory channels, pated work practices. The studies also and tele-work. They also tended to work but many professionals also had visual provided information about the most more in the evenings and weekends. Both co-ordination aids that support aware- important patterns for communication, groups organised their work at home and ness of team status, such as work pro- co-ordination and co-operation. The at the office in very similar ways – most cess charts, geographical distribution results from mobile workers showed that of the working time was devoted to com- charts and maps. Team members de- this user group mostly worked syn- puter work and asynchronous communi- sired resources, which could be used chronously, either face-to-face within cation, i.e. e-mail and fax. instantly to solve time constrained teams or via various audio communica- tasks, especially visual and auditory tion systems that support awareness of An unexpected finding was that, even mobile access to information to verify colleagues’ geographical position, status though management tended to be positive the team status. of work processes, etc., but working dis- to telecommuting, the colleagues of tributed from other functions in the telecommuters were often critical, fear- The authors also conclude that informal organisation. For all user groups, the pre- ing isolation and complications in their communication in distributed teams is sent mobile communication was regarded own work as a result of introducing not well supported by extant tech- to be the most important working tool, telecommuting in the organisation. nologies. Identifying and dealing with since the communication system is the such factors is necessary for the success- only tool that supports real-time access to The authors conclude that the tests had ful implementation of future multimedia other functions in the organisation. revealed two important categories of services for mobile professionals. functionality in the five professional groups studied (see Figure 13):

CONSTRUCTION WORKERS INFORMATION CONSUMPTION

FIRE FIGHTERS

INFORMATION INFORMATION PRODUCTION

PLUMBERS

INSTANT VISUAL INSTANT TEAM STATUS TEAM STATUS SALES REPRESENTATIVES AUDITORY INSTANT TEAM STATUS

TV REPORTERS

Figure 13 The diagram shows the relationships between User-groups and Functionality in their current work-situation (after [57])

Telektronikk 3/4.1997 47 5.2 Collaborative tools in dis- support which roles and tasks. They also 5. Joint Editing tributed research projects wished to know the exact requirements 6. Joint Viewing for communication in a distributed re- 7. Message Conferencing International research projects require a search project. 8. Scheduler high degree of communication and col- 9. Shared Blackboard, and laboration over long distances, and work The authors dealt with this by distri- 10. Workflow Management often has to be done to tight time buting two questionnaires among the 3. Inquiry as to whether specific combi- schedules. Efficient communication members of R&D departments in the nations of tools addressed under Part 2 systems have thus become a prerequisite EURESCOM (European Institute for have synergistic effects, and expected in such work. In what way synchronous Research and Strategic Studies in use of tools for specific classes of col- and asynchronous multimedia communi- Telecommunications) community. The laborative work. cation tools can support work in distri- aim of the first questionnaire was to buted research teams was studied by obtain information on: One important finding was that nearly all Böhm*, Corte#, Marion#, Oberndorfer*, • What kind of telecommunication activ- respondents used the telephone, e-mail Schmitz* & Uellner* [58] from ities do users carry out in the context and face-to-face meetings almost daily. *Deutsche Telekom, Technology Centre, of their EURESCOM work Fax, file transfer, postal mail, exchange Darmstadt, Germany, and #Telecom of diskettes, Internet and bulletin board Italia, CSELT, Torino, Italy. They find • What kind of computer support is was used weekly or less frequently, while that even e-mail (if available) seems to available to users today video telephony, video conferencing, be insufficient for the exchange of • What kind of computer support do videotext, voice mail and computer-sup- complex information, and face-to-face users think would help them with their ported telephony was hardly ever used. meetings are too costly in terms of travel daily communication problems. time and expenses. However, multimedia The authors’ main findings were that collaboration tools, if properly chosen, The second questionnaire had three parts over three times the current number of seem to be a solution to these problems. asking information about: users expect to be using Joint Viewing and Joint Editing tools in the future, Since project members play different 1. Personal data and characteristics of the although expected frequency of use was roles in a project, such as project partici- EURESCOM projects people were not expected to change significantly. pant, task leader and project leader, they involved in Twice the current number of users also perform different tasks, and a 2. Current and expected future use of expects to use Shared Blackboard, Work- number of different activity classes were collaboration tools, classified into the flow Management, Decision Support and identified, such as work process, decision following 10 groups: Desktop Conferencing tools. Respon- making, design, discussion and co-ordi- dents estimated that they will use Shared nation. Figure 14 gives an overview of 1. Decision Support Blackboard and Desktop Conferencing the interrelations between roles and tools. 2. Desktop Conferencing tools more often in the future, while The authors wished to determine which 3. Document Management Workflow Management and Decision capabilities of collaboration tools could 4. File exchange Support tools are expected to be used less frequently. The number of users of Message Conferencing, File exchange, Document Management systems and Scheduler tools is not expected to change significantly in the future, and the frequency of use of Scheduler tools is even Tool Super Class support expected to fall (see Figure 15). Roles (Tool Family) In conclusion, the authors remark that members of distributed research and development teams require and are willing to use collaborative tools in order to consist of possess perform define improve the effectiveness of their joint work efforts. The tool classes that will be of greatest importance are Desktop Con- Tool Activity ferencing, Joint Editing, Joint Viewing, Class Classes Shared Blackboard and Workflow Management. The authors find the respondents’ modest confidence in Docu- possess consist of ment Management tools a bit surprising. support Obviously, many users are satisfied with Capabilities Activities their e-mail and File Transfer tools and they do not see any urgent need for alternative systems. Figure 14 The diagram shows the relationships between Roles and Tools in the distributed research project (after [58])

48 Telektronikk 3/4.1997 5.3 Virtual office environment 300 with shared room awareness Joint Viewing

One advantage of working from a home office is to be able to work in seclusion Joint Editing and not be constantly observed by col- 200 leagues, while this same isolation when working at home, away from colleagues, Workflow Management is often quoted as the main drawback of working at home. How can the tele-com- Shared Blackboard muter have the best of both worlds, i.e. both to revel in total seclusion when that Decision Support 100 is most appropriate and to enjoy the Desktop Conferencing social awareness of colleagues when that is more desirable? Change in number of users [%] Scheduler To be able to strike a balance between Document Management these two incompatible wishes, Honda, Tomioka, Kimura, Ohsawa, Okada & 0 Matsushita [59] of Keio University, File Exchange Message Conferencing Yokohama, Japan, have realised a system which enables home workers to have a -50 home office in 3D virtual space and work -100 -50 0 50 100 150 in it together with their virtual colleagues. The system provides a virtual, shared Change in frequency of use [%] room on a Sun station and provides simu- Figure 15 Expected changes of usage. The expected percent change in the number of lated awareness of other colleagues to aid users (ordinate) is plotted against expected percent change in frequency of use communication. In a real office, we are (abscissa) for ten common collaborative tools applications (after [58]) able to sense other colleagues’ awareness of us, such as actions and gaze, and other occurrences in real time while working.

To realise awareness, the home office was provided with two interfaces called such as speech, and involuntary, such as The authors conclude with an informal Around View and Sound Effect. In the sneezing, and environmental sounds such evaluation of the system on the following real world we are able to sense if some- as the movement of people around you, five points: one next to us sits down or stands up, chairs being moved, office equipment even though we do not look directly at 1. Presence of the user such as printers and copiers, and back- the person. This is because our field of ground noise such as ventilation systems, 2. Presence of the other workers vision extends more than 180°, while the traffic, etc. With Sound Effect added, field of vision where we see details 3. Facility of communication users working at home should get a more extends only about 60°. The Around realistic awareness that other people 4. Position awareness View interface is based on the human actually work in the office. visual field, i.e. the field of vision sub- 5. User interface. tends 180°, but only the middle 60° The study consisted of a number of tests, depicts any detail while the outer flanks The authors show that the system pro- involving many different methods, in- gradually show less and less detail. With vides the expected awareness level for cluding video filming and recording of Around View users can therefore have the task at hand. When little ‘social con- EEG (electro-encephalography) brain awareness of a person’s motion in the tact’ is required while working on a diffi- waves to determine degree of intellectual side fields but detailed picture of a cult task the system automatically re- concentration from viewing the video person sitting in front of them. To see duces awareness level to Narrow, but it and analysing the EEG for presence of someone at the side in any detail, the user returns to Normal awareness in an ‘ordi- beta-waves. The size of the virtual has to turn, but this information is shared nary’ work situation and changes to Wide awareness space could be varied in three by the other office workers, who will be awareness when it senses that more levels, Narrow, Normal and Wide. The aware that they are now being seen. A ‘social contact’ is preferred. level was determined by and varied sensor on the home worker’s swivel chair inversely with degree of concentration, sensed the rotating angle, and if this the more the worker concentrated on the 5.4 Tele-working in Rome: exceeded 30° the view of the office also task at hand, the smaller the awareness a case study rotated to the same side. space (see Figure 16). People in the city of Rome spend an In the real world we also have acoustic average of three weeks or more a year awareness information. This consists commuting to work. Most of these both of human utterances, both voluntary

Telektronikk 3/4.1997 49 Unconcentrated Concentrated

Awareness Space (Providing Audio/Video)

Figure 16 The experience of shared-room awareness space changes according to the tele-worker’s mindset: When the concentration on a task increases the perceived awareness space shrinks to the proximal personal area (right panel), but if the mental concentration diminishes the perceived awareness space expands to encompass the whole room (left panel) (after [59])

people are ‘information’ workers. This view. He finds that it is just as important the centre to the periphery, or outskirts, begs the classical question: “why do to focus on the suitability of products as of Rome tele-worked significantly more these people have to go to an office 30 or it is on process, performance criteria, often (average 7.3 days/month) than more kilometres away to work on a per- productivity or establishing trust. On the other groups, while people who com- sonal computer or talk to a telephone that basis of initial ‘scouting’, 35 employees muted similar distances from the out- could be in their homes or in an office from ten different municipal units in the skirts to the centre chose to tele-work much closer to their homes?” [60,61]. centre of Rome were selected to partici- less often (average of 4.1 days/month). pate in a project lasting three months. In Rome, with nearly three million in- A flexible tele-work scheme allowed the The overall number of interpersonal habitants, some 5,800,000 commuter participants to work either from their communications decreased with tele- trips, average length of 10 km, are made homes or from a local tele-working working from an average of 15.9 to an every working day. Preliminary results centre. Data was collected via question- average of 12.7. The number of commu- from field trials with tele-working in the naires administered after each tele-work- nications with superiors and collabora- municipality of Rome, the Rome Traffic ing day and after two or more traditional tors remained virtually unchanged in Decongestion Tele-working Project office days. tele-working. While the number of com- (Rome TRADE), is reported by munications with colleagues and clients Matarazzo [62] from Fondazione Ugo Based on a sample of only 18 partici- decreased, the number of communica- Bordoni, Rome, Italy. The main objec- pants, some preliminary results are re- tions with friends and relatives increased. tives of the Rome TRADE project were: ported. The average frequency of tele- to identify characteristic commuter ori- working was about four days per month. All tele-workers rated the tele-working gin/destination patterns; to carry out tele- The number and length of commuting situation as more interesting, more work trials among employees of the trips did decrease with tele-working, pleasant, more useful and more pro- Rome Municipality; and to develop a from 3.7 trips/non-tele-working day of ductive than the office situation. The scenario for introducing tele-work in average 30 km each, to 1.6 trips/tele- technology caused few problems: Most other public institutions. working day of average 25 km each. The tele-workers communicated by tele- results show no significant effect of phone, fax or computer e-mail, and used Although the basic rationale for the pro- gender, age or profession on the fre- a computer for most of their work either ject was to find ways to decrease traffic quency of tele-working. However, a most at home or at the work-centre. congestion in Rome, the author initially striking effect of the direction of com- stressed the importance of planning tele- muting on the frequency of tele-working Autonomy in performing the job tasks work from a human factors point of was found: People who commuted from increases significantly with tele-working,

50 Telektronikk 3/4.1997 the tele-workers become more indepen- • Suitable video signal switching pro- 6 Product development dent and ask less for advice, so even cedures though tele-workers interact less with The development process of telecommu- • A ‘multimedia chair’ for the tutor. superiors and collaborators, their deci- nication products and services has sion making is not impeded. changed substantially during the latter Nine students (aged 21–25) attending a years due to the application of various course in Communication Mass Sociol- Finally, Matarazzo reports that the feel- design strategies and the gradual intro- ogy from the University of Rome “La ing of isolation in tele-workers was not duction of various human factors pro- Sapienza” served as subjects. The course significantly changed in the tele-working cedures. These include such evaluating consisted of six lectures given by tutors situation. This runs counter to conven- techniques as user participation, paper using the ‘multimedia chair’. The tional wisdom that isolation, both prototyping, software prototyping, students were distributed in three rooms socially from colleagues or as lack of heuristics, ‘Wizard of Oz’ simulation, that communicated with the tutors’ room. information about what is going on in the cognitive walkthroughs, to mention only After each lecture, a questionnaire was organisation, is one of the most often a few current methods. One user inter- issued and the subjects were asked to rate proffered reasons against tele-working. face that has gained wide acceptance is various aspects of the learning climate, The finding may partially be explained the Graphical User Interface (GUI) com- various usability factors and finally, to by the relatively low frequency of tele- bined with direct object manipulation, give an overall evaluation. working in this study, but the preliminary i.e. clicking and dragging icons to per- results from the Rome TRADE project form various actions that were previously The results reported by Papa & do dispute widely held beliefs among achieved by text input and short com- Spedaletti show that the subjective rat- opponents of tele-working. mands. A well-designed GUI may sub- ings of the learning climate, although stantially enhance the usability of ITC they were generally high, did not change products or services, but just applying 5.5 Broad-band technologies in significantly from the first to the last lec- GUI indiscriminately does not neces- tele-education ture. The usability ratings were also gen- sarily guarantee a user-friendly product. erally high, and did not improve signifi- It is still necessary to work through the Tele-education, which is a special form cantly over the six lectures. The same various design process stages in a sys- of distance education, shares many was also the case with the overall ratings tematic manner and perform the required features with tele-working: i.e. geo- by the students, no significant improve- analyses and, not least, act on the results graphically distributed people working ment, but ratings were generally high. obtained to achieve a better product. from their homes interacting with tutors via telecommunication equipment and The teachers judged positively the pro- services. The main perceived barriers to a cedures centred on their needs, and the 6.1 Clever GUI design gives broader acceptance of tele-learning and teachers designed their lectures using value for the customer tele-teaching are related to various prob- most of the opportunities offered by the lems in interacting with technologies and multimedia systems. The availability of The use of the Graphical User Interface to general resistance in teachers and stu- different multimedia does change the (GUI) has spread rapidly from the world dents alike, especially in the fields of the tutors’ use of the media. There were of the high performance computer to a humanities and the social sciences. Dif- stronger time constraints in a virtual host of products and services in the field ferent virtual classroom situations may classroom than in ordinary face-to-face of telecommunications, and several other be created to suit different topics and the tuition. A reduced sense of presence areas. The same way a good GUI can different needs of pupils, from the simple among students compared to face-to-face enhance the usability of a product, a use of e-mail to the use of very sophisti- situations was perceived by the tutors, poorly designed GUI may be no better or cated multimedia services using speech, and a virtual classroom demands de- even worse than the older method it replaces. music text, drawings, animation and veloping new skills by tutors, e.g. being video to create so-called ‘tele-presence’ able to deliver a lecture in front of a camera. Fleet, Talbot & Winlow [64] from to support contact between teachers and Nortel’s Corporate Design Group, pupils. In conclusion, the results indicate that no Ottawa, Canada, have developed a six- real differences were found in the stu- phase process for the design and evalua- To investigate the effect of various multi- dents’ evaluation of the usability dimen- tion of product interfaces with the clear media systems on the learning climate in sions over the duration of the experiment, objective of offering customers supreme tele-education, Papa & Spedaletti [63] although the system was generally posi- value. Each phase in the process is de- from Fondazione Ugo Bordini, Rome, tively evaluated. In general, the interac- signed to address a critical interval in the Italy, carried out experiments using a tions permitted by the system were satis- product development cycle and contri- broadband network to emulate virtual factory. This kind of experiments are bute to design excellence (see Figure 17). classroom situations in the laboratory. much too short and limited in scope to give any real evaluation of tele-educa- The first phase is a Domain analysis, or In configuring the virtual classroom mul- tion, classroom presence and suitability to understand the problem space, i.e. the timedia system, some important features of multimedia systems in tuition, but they who, what, where, when and why of the were aimed at: can pinpoint specific problems or fea- product domain. To clarify such ques- tures that can be used to improve the tions, various techniques such as site vis- • Possibility of self-view design of future trials. its, interviews, direct observation, sur- • Continuous audio presence veys and questionnaires can be used to identify significant aspects such as: • Continuous video feedback to the tutor

Telektronikk 3/4.1997 51 stakeholders. Prototypes can offer de- adequate understanding of customer 1 Domain analysis velopers detailed feedback in a timely, requirements. Thus, the real measure of cost-effective manner. The authors de- success of any design approach is that it 2 Concept exploration scribe three levels of rapid prototyping: substantially reduces the risks to the manufacturer while increasing the value • Paper prototype; using pencil 3 Concept prototyping to the customer. sketches, bubble diagrams or 4 Concept verification flowcharts to portray tasks, object models and task flows 6.2 Individual differences in design of automated • Walk-through prototype; using anima- services tion, HyperCard™ stacks or Quick- Time™ films to demonstrate func- How can individual differences in users’ tionality performance best be assayed and incor- Product • Software prototype, fully or partially porated in the development of new ser- functional prototype used for usability vices? This question was addressed by testing, demonstrations and design Love, Foster & Jack [65], from Centre of 5 Design refinement transfer. Communication Interface Research, Uni- versity of Edinburgh, Scotland. They 6 Product support The purpose of the next phase, Concept employed a three-step model proposed verification, is to minimise costs in terms by Egan & Gomez [66]. Step 1 ‘assays’ Figure 17 The six main phases of the of money and time by verifying concepts the effect of individual differences such design process. The first four phases may at an early stage rather than at later as age, personality or cognitive skills on be iterated as often as necessary stages in the development. Usually, a task performance. Step 2 ‘isolates’ where (after [64]) number of iterations are required, and salient differences identified in Step 1 these are more cost-effective to perform affect interaction most. In Step 3 features as early as possible. Some common which have been isolated in Step 2 are testing techniques are: ‘accommodated’. Using Steps 1 and 2, the authors performed an experiment to • Focus groups, good for collecting investigate the hypothesis that individual qualitative feedback from users differences in subjects’ performance can • One-on-one interviews, useful for be attributed to individual characteristics, looking at concepts in more depth and that these differences can be isolated to specific sub-tasks in the user-interaction. • current and future market trends and • Usability testing, effective with suffi- needs that may influence the interface ciently developed concepts To test the hypothesis an automated • current ‘best-of-class’ interfaces and • Co-discovery learning, a preferred music catalogue telephone service with their perceived customer value technique for testing interactive menu selection was designed. The cata- products. logue was hierarchically organised with • distribution channel variables that may menus of three to five items. The menus influence the interface By the fifth phase, Design refinement, were arranged like the grouping of CDs • end-user profiles, tasks and require- the design is essentially complete. Fur- in a typical music store: The first level ments ther usability testing for fine-tuning the category menu offered Blues, Jazz, Clas- concept may still be done and missing sical, Rock & Pop, and Folk, the next • system requirements, features and details can be filled in, but at this stage it artist menu level offered a choice of four development schedule is relatively expensive and technically artists or composers for each category, • current and future user interface tech- difficult to make any significant technical the next album menu level offered a nologies and techniques. changes. choice of three albums from the chosen artist/composer, and the last track menu The Concept exploration phase is meant The final Product support phase is meant level offered a selection of three tracks to generate ideas that may lead to fresh to ensure that maximum benefit is de- from the chosen album. ‘Audio-on- user interface concepts. In this phase it is rived from field trials and early product demand’ was provided at the track menu essential to encourage unorthodox think- feedback. It is important to verify if the level, allowing the subjects to listen for ing. Basic techniques in this phase may original design hypotheses were correct, up to 30 seconds to any track from the include: which specific design features were most selected album. successful, and to seek recommendations • Brainstorming to generate fresh con- for improving future products. Thirty subjects balanced for gender par- cepts ticipated. The subjects interacted with the • Organisation and grouping of concepts In conclusion, the authors present an on-line music catalogue via a telephone example of how the six-phase design keypad. The task was to compile a ‘per- • Voting between the best concepts. process was successfully employed to sonalised’ CD by ordering four tracks, design a network managing system. two of which were specified in the In the third phase, Concept prototyping, Their studies have shown that over 50 % instructions and two to be selected by the concrete examples are used to demon- of product defects can be traced to in- subject. Each subject attended four ses- strate the value of the concept to the

52 Telektronikk 3/4.1997 sions, in which a new CD was compiled 1.0 and one each of four psychometric tests was completed: High Verbal Ability 1. NEO – Personality Inventory Revised 0.8 [67], measures neuroticism, extraver- sion, openness, agreeableness and con- Low Verbal Ability scientiousness 0.6 2. AH4 Test of General Intelligence [68], assesses general verbal, numerical and spatial reasoning skills 0.4

3. The Paced Auditory Serial Addition Mean Interrupt Rate Task [69], measures information pro- 0.2 cessing capacity 4. The Rey Auditory Verbal Learning Test [70], assesses both short- and 0 long-term memory. Trial 1 Trial 2 Trial 3 Trial 4 Experimental Trials Two objective performance measures were also obtained – the interrupt rate Figure 18 Mean Interrupt Rate (ordinate) is plotted as function of trial order and the silence rate. Interrupt rate is the (abscissa) for two different verbal ability groups: High Verbal Ability (continuous ratio of the number of key-presses before line) and Low Verbal Ability (broken line). The differences between the groups were the end of a system prompt to the overall significant ( p < 0.001) (after [65]) number of key-presses made by a subject to menu prompts during an experimental trial. A ‘silence’ was registered when a response was not made before a ‘time- 1.0 out’ period of 5 seconds after a system prompt. The silence rate is the ratio of number of ‘silences’ to the overall 0.8 number of responses made during an experimental trial. 18 - 38 years

No significant differences in silence rate 0.6 were found for any individual character- 39 - 59 years istics, including age and gender. Also, no significant differences in interrupt rate 0.4 were found for personality, memory or processing ability factors. However, sig- Mean Interrupt Rate 60 + years nificant (p < 0.001) effects of age and 0.2 verbal ability on the interrupt rate were found (see Figures 18 and 19). The results suggest that differences in 0 user behaviour with an automated service Trial 1 Trial 2 Trial 3 Trial 4 may be explained in terms of individual Experimental Trials differences. Also, the experiment has shown that once the salient individual Figure 19 Mean Interrupt Rate (ordinate) is plotted as function of trial order differences have been identified, it is (abscissa) for three age groups: 18–38 years (broken line), 39–59 years (dotted line) possible to determine where these indi- and 60+ years (continuous line). The differences between the groups are significant vidual traits produce their greatest effect. (p < 0.001) (after [65]) Since it was found that age and verbal ability contributed significantly to the variation, the next important issue to be addressed would be how to ‘accommodate’ these differences (Step 3 [66]). One questionnaires, and laboratory tests. approach would be to use speech input 6.3 Fast & facile feedback in These are both costly and time-con- rather than key-presses. According to the intranet-working suming to carry out. authors, several of the older subjects actually expressed that they would have The traditional methods for obtaining Gustafsson [71] of Ericsson Develop- preferred to speak to the system. feedback from users during the product ment, Sweden, points out that the tradi- development phase have been face-to- tional methods, with reply cards, or ques- face interviews, on-site observations, tionnaires do not work well and that the

Telektronikk 3/4.1997 53 was temporarily abandoned since users Send a message to Webmaster could come to expect that this function Please enter you message below and send it to the Webmaser. key would work with all applications. Instead, the Mail-order Help function was implemented as a help menu option. From: I When selected a simple form is presented asking the user to fill in any questions or Your E-Mail Address: I comments and then simply select the screen button Your Message: (Figure 20) to transmit the user feedback I to the appropriate recipient (no e-mail addresses needed, no alternative check boxes to be selected, etc.). In order to provide prompt replies, the e-mails were received by an e-mail alias and redis- tributed to the members of the development team. The team member who replied also had the reply distributed Click here to Send your MessageClick here to Erase your Message automatically to the other members of the team so that they could know how the feedback had been dealt with. Figure 20 An example of a simple, well designed, web-based feedback tool. Informal, When a particular Mail-order Help feed- candid and impromptu comments can easily and quickly be sent to the developers’ back was considered to be of more gen- newsgroup (after [71]) eral interest, the question or comment was added, along with the reply, to a newsgroup in order to make the information available to the whole development Announcements group. However, the name of the person 2% who made the feedback was deleted and Comments replaced by “Someone just asked …”, 8% permitting users anonymity to ask ‘silly’ questions (Figure 21). Depending on the confidentiality of the development project, feedback and answers can be shared in a larger newsgroup within the organisation, or even outside over Internet.

Gustafsson found that substantial savings could be realised when the help menu option and newsgroup were used in the Answers Questions development work on an operation & 44% 46% maintenance product (OMGR Object Manager for use with Ericsson’s ATM Broadband System).

In conclusion, Gustafsson sums up his findings with the e-mail method as easy to use, cost-efficient and much appreciated by the users. Little effort was required to deal with the Mail-order Help and the newsgroup, and the benefits Figure 21 The distribution of news categories (N = 128) received over one year by the were: designers’ newsgroup. Questions (46 %) and answers (44 %) make up the dominating majority of inputs (after [71]) • Suggested changes got early feedback from users and designers • Feedback from users could result in fast improvements. most obvious feedback systems, intranet The Mail-order Help is a function that or internet e-mail, have not been properly was originally conceived as a “What-the- Another valuable feature of a newsgroup utilised for this purpose. He proposes a heck?!” dedicated function key. The idea is that it acts as its own tutorial: Begin- system called Mail-order Help, which is was to be literally only one keystroke ners may start by reading what the more installed with the software to be assessed. away from feedback help, but this idea seasoned developers have written, and

54 Telektronikk 3/4.1997 when they feel more confident they may play (MUD) that continually displays the Ling & Hareland divide the comments start submitting their own replies and cumulative costs of an outgoing call to received in the study of metering unit comments. As an alternative to more the caller. 100 other residents were pro- displays into two categories, positive and traditional methods, the use of the vided with a cordless telephone. The negative. The positive comments re- intranet and Internet for e-mail and news- hypothesis was that providing technology volved mainly around three themes; groups may have much to recommend it. that displays the consumption of meter- monitoring costs, saving money, and saving units or that allows higher avail- ing time. The negative comments centred ability and making calls in more comfort around the notion that increased cost 7 New services, user would in some way influence the tele- awareness inhibits spontaneous telephone requirements and phone use. use which affects the social nature of telephone conversations – it makes one marketing The results should be a quantification of think twice before making an ‘unneces- a series of qualitative questions regarding sary’ (i.e. social) call. This kind of com- The proliferation of new services and the respondents’ immediate impressions ments underscore the importance of the equipment is accelerating beyond all of the two types of telephone equipment. telephone for holding friends and rela- expectations as the result of very aggres- Before being provided with the new tives together. sive marketing, while knowledge of user equipment, each of the two respondent requirements and reactions to the new groups had to give their immediate These findings conflict with the basic services are lagging behind. Some new opinions as to the positive and negative hypothesis of the study, i.e. that accurate products and services like GSM cellular aspects of the metering unit display and information on the costs of telephone use mobile telephones and the Internet have the cordless telephone, respectively. would result in more telephone use since been taken up despite often inferior user most users actually overestimate the cost interfaces. This may be because it was The authors’ findings show that these of telephone conversations. felt that the products fulfilled some technologies stimulate a mirror pattern of important needs and functions, i.e. the associations. The negative aspects of The comments received on cordless tele- perceived status, freedom and mobility cordless telephones, i.e. that increased phones were also divided into positive offered by the mobile telephones, or by comfort and availability would lead to and negative. The positive responses fell the huge amount of information available increased use of the telephone, increasing mainly into two broad groups, dealing on the Internet and the relatively low costs, were mirrored by the positive with mobility and with convenience. The price of accessing it. aspects of the metering unit display, i.e. most obvious comments were that cord- that making the cost more prominent less telephones allow users a much Still, it is a fact that many people feel would lead to less use of the telephone, broader range of mobility, both indoors estranged by the new sophisticated tech- reducing costs. And conversely, the posi- and out of doors, which fixed telephones niques, often requiring computer literacy tive aspects of cordless telephones, i.e. can not offer. Typical comments were or advanced knowledge of complex user that increased comfort and availability that it allowed more privacy, parallel interfaces to be utilised. Even the lovely would lead to increased use of the tele- tasks, and not least, less risk of missing telephone system is now pervaded by phone and more social interaction, was incoming calls; i.e. avoiding having to interactive speech-based services where mirrored by the negative aspects of the rush to answer a ringing telephone. the users have to dial their way through metering unit display, i.e. that showing While parallel tasks are tolerated, they several levels of menus even for very the cost of using the telephone more are often seen as breach of good form. simple tasks such as calling a taxi or prominently would lead to reduced use of Social norms dictate that a certain ordering a cinema ticket. the telephone and less social interaction. amount of attention should be given to the conversation. If background noises 7.1 User expectations of The users’ metaphors for telephone use (e.g. children bathing or kitchen sounds) metering unit displays also have a gendered difference; the idea become too intrusive, the caller may and cordless telephones of the telephone as a utilitarian device, come to question the attention given to mainly maintained by men, or the tele- the conversation and the legitimacy of In a domestic setting, does having the phone as a social device, mainly main- the interaction. actual cost of using the telephone contained by women. tinuously displayed have any positive or The negative comments on cordless tele- negative effects on the use of the tele- The fundamental tone in the comments phones fell into two groups; those deal- phone, and does the use of cordless tele- reported by Ling & Hareland is a deeply ing with user behaviour and those deal- phones have any positive or negative held notion that the telephone use is ing with the characteristics of the cord- effects on the time of telephone use? Or expensive. This finding is quite certainly less telephone. The issue that received in other words, is there a price of being due to the fact that up to the early 80s the most concern was the fear that the conve- social? unit cost of making a call in Norway was nience of using cordless telephones one of the highest in Europe, and tele- would lead to increased telephone use. These questions were addressed as part phones in Norway were, in effect, According to the authors, this is similar of a larger project by Ling & Hareland rationed (people had to wait up to two to the reasons given by the Amish people [72] from Telenor Research and De- years to receive a telephone subscription, in the USA. Although they have chosen velopment, Norway, in a survey con- and children actually took over their par- to accept the telephone, they do not ducted among 200 residents in the Bryne ents’ telephone subscription when par- encourage its use for ‘idle’ gossip on area in southern Norway. 100 residents ents died and did not report the change to moral grounds and try to restrict its use were provided with a metering unit dis- the operator). by not having it in their homes, but

Telektronikk 3/4.1997 55 placing it inconveniently in an out build- • Technology monitoring: sity of Technology, Finland. They have ing. Although the Norwegian rationale - Early recognition developed a classification of extant was economic rather than moral, the - Discovering and assessing innova- telecommunication services that is based notion that greater comfort provided by a tions on user communication task analysis. As cordless telephone would lead to more - Transfer innovations into develop- a case they first analysed the communi- calls was the same. Some other practical ment scenarios cation needs and problems of two mobile issues were concerns that the cordless care-work organisations. They then • Co-operation: handset would be easy to misplace and analysed existing telecommunication - Group management that users would become dependent on services and grouped them to meet the - R&D strategy the device, and, not least, that short specific needs of the organisations. This - Select experts. battery life would be a problem. These was also carried out for single users. findings provide insights into the folk The needs and uses of telecommunica- definition of telephony, and especially Mobile care workers calling at the tions depend upon a number of inter- the tension between the consideration for clients’ houses were observed during dependent socio-cultural, political, eco- cost and sociability is set into relief. It is work. 20 workers were interviewed about nomic and technological factors. Aggre- against this background that new tech- their communication with their organisa- gations of general influences, develop- nologies must be domesticated. tions and participated in simulated comments and trends pointing in the same munication with a group of six home care direction Schyguda calls megatrends. To workers. The data were then used as the 7.3 Societal trends and the illustrate the object of the study he gives basis for grouping existing telecommuni- telecom markets an example. cation services according to the needs of the home care organisations. A similar, The ongoing transition from an industrial A most significant demographic mega- but more general method was used to to an information society will ultimately trend is that the percentage of senior analyse the communication needs of 15 affect most areas of life. The conver- citizens in the populations of the industri- single individuals from all walks of life gence of media, computers and telecom- alised countries will grow rapidly during and tabulate the findings. munications made possible by the new the next few years. There are currently ICT technologies is at the core of this 16 million elderly people in the Federal The authors find that the current trends of development. These changes will have Republic of Germany, which comes to service personalisation and service inte- significant effects on work, leisure, con- about 20 % of the total population, and gration would provide good solutions for sumer behaviour and social life. On the this is expected to rise to at least 25 % by service selection problems and message one hand, this will lead to more inte- the end of this century. The character- chaining problems for most of the grated systems, on the other hand it will istics of the target group can be used as a respondents. As an example the authors lead to world-wide networking. As a basis for inferring what needs must be mention that a message left using a par- result of these changes and the opening taken into consideration by the providers ticular service, e.g. voice message or e- up of the European telecommunications of telecommunication services and mail could be picked up by any service market in 1998, operators are faced with equipment. In the case of senior citizens, independent of the service used for send- the challenge of defending their existing this would cover such characteristics as ing the message. In conclusion, the markets and to develop new market reduced mobility, high demands on qual- authors warn about mixing up informa- sectors by introducing innovative tech- ity, strong family ties, declining fitness tion about user communication needs and nologies and services. and more available time. Other socio- marketing information on estimated use cultural megatrends would be the ‘gener- of new services. To meet these challenges, Schyguda, [73] ation X syndrome’ or the ‘Nintendo kids’ from the Technology Centre, Deutsche craze. Telekom, Germany, studied the in- 7.5 Introducing new telecom fluences of various trends. Using the In conclusion, Schyguda finds that the services WWW-Online-Expert forum set up by methods of ‘trend gurus’ such as Naisbit, Deutsche Telekom, societal, economic Popcorn, Horx & Gerken are often vague Creating and introducing new telecom- and individual trends in Germany and on methodology, use ambiguous termi- munication services and products in a Europe were extrapolated to gather infor- nology, lack substantiation and source rapidly re-regulated telecommunication mation about future markets. For such references, contain imprecise information market pose some special problems. work the author proposes interlinking on trend diffusion over time, and contain Svendsen, Stenvold, Folkow & Akselsen four core skills: too many inconsistencies. Therefore, it is [75] from Telenor R&D, Norway, have important to develop new techniques. devised a method for creating and intro- • Market development potential: ducing new telecommunication services - Socio-economic environment of and products. product planning 7.4 Matching telecom services - Initiation of markets with user needs The authors first analysed a number of - Issues of user acceptance R&D projects for promoting new - ICT-technologies and user be- A very central question is; are the struc- telecommunication services and products haviour tures of telecommunication services such as tele-medicine, virtual corpora- based on an overall view of the user • Methodology: tions, tele-work, distance education, communication needs? This question is - Analyses of the future communication within local public central in the work reported by Ranta- - Workshops sector, and monitoring the environment. aho & Leppinen [74], Helsinki Univer-

56 Telektronikk 3/4.1997 In analysing the results, the authors opened the way for Internet, e-mail and were on holiday, a burglar who calls found that the activities that produce the WWW, integrating the mobile tele- would be met by an answered phone and deliverables leading to project success phone with lap-top computers has given mistakenly believe that the home was fell into four categories: us the portable office, the list may be inhabited and go somewhere else. This extended. Integrating microprocessors use of this supplementary service became • Creating examples into different equipment, such as PBXes so popular, especially during Easter and • Mobilising opinion leaders and terminals, will give us many new Christmas holidays, that the capacity was services and features that were quite exceeded and many prospective users • Exposing obstacles unthinkable until recently. Intelligence in were met by the message that the service • Acquiring knowledge. the network will be able to keep track of was unavailable. all subscribers and route calls according In analysing the effects of projects to to urgency and personal priorities and This is just one example of a simple ser- create, promote and introduce new tele- agendas. Hopefully, such systems may vice, which at the outset had no more communication services and products, one day serve us as a loyal, but now potential than many other supplementary the authors highlighted two problems: nearly extinct, secretary or butler – for services, but that was taken up because some reason robots in science fiction of a life-style and concerns about being • Services are becoming more complex films all seem to speak the distinct and burgled. We will see many other such and are more integrated into peoples’ correct English of a well trained butler. success stories in the future, but we will lives not be able to know beforehand which • A service must reach some critical The problem today is that most of the service will be a hit. One important task mass before being useful. sophisticated new possibilities are under- will therefore be to monitor the life-style used; though many people agreed that and actual needs of subscribers so new These problems were addressed by: they wanted them when they were asked. trends can be assessed as soon as they Often only one or two new products seem appear. Operators who do not see the • Create real-life examples to reduce to fill a need and are adopted by any trends and react to them in time may lose complexity for the user number of users, often despite bad usabil- much revenue. • Mobilise opinion leaders as promoters ity (cf. the early video recorders), while to reach critical mass the majority of new features are seldom, With the convergence and integration of if ever used. The question, then, is how to different services and media, we are • Disclose any obstacles. make all these wonderful new services going to see some very spectacular new and features so simple and intuitive to use services, but which services will make it The authors find this method useful for that people will actually use them. in the marketplace and which services projects involving the following prob- will be failures cannot be determined a lems: This of course is the domain of human priori – new trends are never easy to pre- • Users need help to see how to adapt factors, and here lies a challenge that is dict. Sometimes even the most unlikely services to their needs different from the traditional human fac- products become hits, cf. the Walkman® tors in telecommunications. One im- or the Tamagoshi®, products that only a • Users do not exploit technical possi- portant source of knowledge to this field year before they appeared would be bilities is through studies of what people actually regarded as unthinkable. • Use of a service depends on others do, what they actually need and how they using the service. redefine services and equipment accord- Human computer interaction (HCI), ing to the changing ways of life. To take graphical user interfaces (GUI), object Thus, there is a clear need to address one example: diverting telephone calls to manipulation interfaces, may ultimately these specific telecommunications prob- another number has been a feature of provide services that are very simple to lems to improve the uptake of new ser- PBXes for many years, you use it if you use. This trend is emerging very fast: vices and technologies. go into another office, but expect an Why use cryptic codes and ungainly pro- important call. When this service was cedures when a graphical user interface introduced in the public network, it was and object manipulation would make the 8 Convergence – integra- given the code: *21*# (and #21# to cancel the ser- puting and the Internet vice). This is not a particularly user- 8.1 ‘Uncountable’ smart friendly code, mainly because there are a features that no-one uses large number of other supplementary ser- The most conspicuous current trend in vice codes to remember. From the start telecommunications is the integration of A large number of new ‘smart’ features the service was not well known and telephony and computers – i.e. the con- in PBXes, to be accessed from terminals therefore not much used. vergence of different media. This by cryptic codes, do not necessarily merging of different media will create a improve usability or provide efficient and However, burglars in Norway have a multitude of conceptually new possi- satisfactory telephone services, and are modus operandi where they phone bilities, possibilities that until recently therefore seldom used. Lindgaard [76] prospective homes to find out if they are belonged to the realm of science fiction. probably put this most bluntly in her inhabited, especially during the Easter Combining telephony and video technol- paper: “Wow – 568 Smart Features on and Christmas holidays. Then the popu- ogy gave us the videotelephone, using Your PABX. What Really Determines lar notion spread that if you diverted your the telephone network with computers the Uptake of Technology?” calls to a neighbour or a friend when you

Telektronikk 3/4.1997 57 Will adding another package of functions achieved and guaranteed in CTI tele- caller, or on how calls should be routed improve the usability of a PBX or a phony, and is generally accepted by the according to personal needs and wishes, smart terminal if the user requirements users, CTI telephony will be the pre- requiring a solid grasp of how the system for representation of services and interac- ferred way to communicate. Until then, worked. Because of the amount and tions are not carefully considered? The we will have to live with a system that complexity of the date to be entered by answer to this is no, according to provides a multitude of smart features the subscribers, the authors chose to Herstad*, Hüttenrauch# & von Niman# that no-one uses because the user inter- employ a graphical user interface acces- [77] from *UNIK, Norway, and #Erics- face is just too bothersome. sible over the Internet rather than the son Infocom Systems, Sweden. They common interactive telephone interface. present findings and conclusions from a 8.2 Your intelligent agent on user interface development process and the web To evaluate the intelligent agent the usability studies with a PC-based graphi- authors recruited eleven computer- cal user interface for integrated computer In order to fully take advantage of the literate employees with no prior knowl- telephony. many advanced new telephony services edge about the agent to fill in a service that soon will be offered, subscribers will registration form presented via a Based on a metaphor of desktop applica- be required to enter information into the Netscape Navigator® browser running on tion-windows and graphically repre- system computer and update the data a networked PC. The evaluators also sented services and following repeated associated with their accounts. To filled in a short questionnaire and dis- cycles of end-user need studies, usability achieve this, the subscribers must be able cussed it with the experimenter. studies, prototyping and software en- to interact with the system computer that gineering iterations, a computer tele- controls the service. This is normally per- A registration took on average 9.7 phony integration (CTI) application formed over a very restrictive communi- minutes to complete. All except one sub- called Personal Screen Call™ was intro- cation channel, usually making use of ject found this acceptable. Although the duced by Ericsson. touch-tone (DTMF) or voice input and tested interface supported only the auditory feedback. This restrictive administration of the service, the GUI The visualisation of available services channel is highly unsatisfactory to most interface may equally well be used to and options without the need for re- users, and this creates a great hindrance control services. The evaluators revealed calling codes or having to look them up to the acceptance of new advanced ser- several flaws in the system; one was that was a clear advantage over traditional vices. What, then, is the preferred manner many users actually forgot to submit the systems, and the system met with much for subscribers to interact with the service data they had entered to the system user approval. The availability of up-to- computers of these advanced systems? because the Submit button was not very date directories, the simplicity to select prominent and easily overlooked. The and click icons to access services was Mané & Rabin [78] of AT&T, Holmdel, authors’ most interesting finding, though, especially well received. NJ, USA, have addressed this question. was that 91 % of the evaluators actually They also encountered this interface preferred to use the web to using pen- However, there were also some import- obstacle when designing the user inter- and-paper to register for the service. ant problems. Computer systems are still face for a telecommunication ‘intelligent not as reliable as the traditional telephone agent’ service. To by-pass the restrictions Mané & Rabin find that they have system, and when the system crashes you imposed by the telephone-based inter- learned some valuable lessons from their loose all your fancy telephone options, face, they chose instead to develop a study: The KISS principle (Keep it even though an ongoing conversation graphical user interface (GUI) that is simple, stupid) is probably the single may not be disconnected. Also, setting available to the subscribers over the most important factor when designing a up a user-friendly and practical system Internet. The authors’ main goal was to user interface. One must remember that can be a problem. Just providing a user explore how ‘intelligent agents’ can for a system to be usable for most sub- with the installation diskettes may not effectively be used to offer subscribers scribers it must also be usable for the less necessarily result in a positive first-time new features and services in the telecom- bright users. Half of all people are less encounter with CTI telephony. munications field. To explore this field than average intelligent (this is an in- they developed a personal assistant for evitable effect of the definition of the As traditional supplementary services are telecommunications (PAT) that offered intelligence scale) but a factor that is eas- getting more complicated, telephone the functionality resembling the services ily overlooked since most engineers, numbers are increasing in length and the normally provided by a telephone recep- designers and programmers are aca- number of available features and options tionist. To be a truly personal assistant, demics, and academics are predomi- in the new systems is growing all the the service must be customised so that nantly recruited from this upper half of time, CTI telephony is certainly going to the ‘agent’ can tailor its services to the the intelligence distribution. Engineers win in the end. However, there will be an needs of each subscriber. The authors and designers usually design for them- imperative demand that the new systems also chose to forego machine learning, selves and their peers, which means that must be at least as secure and safe as which is usually an important attribute of large groups of users have to cope with today’s telephone systems; secure against intelligent agents, and instead make use systems that may often be too difficult crashing, safe against important informa- of user-defined rules that rely on the sub- for them to understand and use. Engi- tion being lost. We have come to trust the scriber supplying the information. This neers and designers may find it difficult traditional telephone system to the point information ranged from simple opera- to put themselves into the position of less that we come to entrust our personal tions, such as furnishing a telephone intelligent users, but by following the health and safety with the telephone. number, to complex specification of the KISS principle they may still come up When this level of security can be different options to be presented to a with usable designs.

58 Telektronikk 3/4.1997 Take the users’ perspective is another 8.3 Is the ‘surfing’ metaphor Learning to use a computer system (or lesson, but as just mentioned, this may be bad for novices? any other complex system) usually in- difficult for some engineers and de- volves constructing mental representa- signers. The problem is that designers The metaphor, i.e. a resemblance, like- tions of some aspects of the system. This have too much knowledge about the ness or similarity, is a powerful tool for mental model is then utilised for predict- system they are designing to place them- conveying information on something ing the system’s behaviour. A model selves apart from it and see it with the completely novel by using well known need not be a ‘correct’ representation of eyes of the user. Nonetheless, this is an similarities. Not least in modern informa- what actually is going on in a system important exercise to perform whenever tion technology we use a number of (few are) to be usable, but a model possible and should not be too difficult metaphors, well-known examples are the should not contain elements that may with some practice. memory metaphor, the desktop metaphor, lead to incorrect actions. A well-chosen and the window metaphor. The World metaphor may induce the formation of a Take advantage of the visual display. A Wide Web is an Internet service for constructive mental model, but few visual display offers the use of a graphi- accessing information and services that is metaphors have a wide enough scope to cal user interface, which is much more growing fast in popularity, not least due encompass all the possibilities, and most effective when dealing with complex to its graphical user interface (GUI) that metaphors quickly run into trouble when information than a monotonous reading allows users to utilise the web without stretched beyond their original meanings. of an auditory menu. For simple interac- knowing the underlying computer net- Many metaphors emanate arbitrarily, tions (e.g. replying ‘yes’ or ‘no’, or mak- work architecture. It is thus possible to often because of some conspicuous, but ing a simple choice between two options) begin ‘surfing’ (another metaphor) the insignificant likeness, and may therefore an auditory prompt may be just as, or web by use of any of the many web convey false assumptions to the user of more efficient. browsers using a graphical user interface how the system (viz. the Internet) func- without any prior training or guidance. tions. According to the author, many Design for incremental learning. The such false assumptions are acquired user rarely needs to know all about a Does the ‘surfing’ metaphor actually through public media. system to use it. It is far better if the user communicate a correct mental user can start to use the system with only model of what is going on? What mental Ranta-aho interviewed ten novice users some initial knowledge, and then learn as models are created by the users from the who were reasonably computer literate he/she goes and new features are taken often arbitrarily chosen metaphors? Is a (5 male and 5 female, aged 20–51 years) up. Nothing is so discouraging as to have well-chosen metaphor conductive to pre- before their first experience in using the to read a thick user manual before being senting the Internet in a manner that does WWW, focusing on their ideas of what able to use a new service. Give the user a not contradict the functioning and system was happening when they used the web. simple start-up instruction and build on architecture, or is any correspondence The interviewees only knew about the common knowledge. between the system and the metaphor web from newspapers, TV and what they superfluous? These questions are add- had heard from friends. They were also Present only task-critical information. ressed by Ranta-aho [79] from Helsinki given a seven question questionnaire (see Only information that is really necessary University of Technology, Finland, who Table 3). In addition twenty persons par- for performing the task at hand should be in an experiment investigated the way ticipating in a WWW course were inter- presented, making the task simpler and users perceive systems and create mental viewed before and after the course on faster to complete. This principle also models. their ideas of what happens during the applies to help information – it is most frustrating getting the whole instruction manual displayed on the screen when asking for help. Table 3 Problem solving questions used in the interviews

Ease of communication and ease of soft- 1) While using the WWW, when you click with the mouse onto a link, what ware distribution in web design. As happens? shown above by Gustafsson [71], the web can actually be a most efficient 2) Is it possible to get some kind of virus using the WWW? vehicle for user interface design through 3) You are currently reading on-line a web page of a foreign university. Suddenly, ease of communication and ease of soft- someone stumbles over your computer’s electricity cable and your computer ware distribution. Using the web actively gets switched off. Will you stay as user in the computer of the foreign university? provides the designer a number of advantages, not least the fast and easy en- 4) You are reading some web pages. Suddenly, your network connection breaks. vironment for collaboration, both with Is it possible for you to still see some of the pages you just read? peer designers and with evaluating users. 5) You are reading a web page, when the author of the page updates it at the same time. Can you see the changes immediately? The authors conclude that the use of graphical user interface was most enthu- 6) You are reading a web page with a form to fill in, and you start filling in the form. siastically received by the users and At the same time, another person starts reading the same page from his com- could also be applied to many other tele- puter. Will he see your entries? phone services besides the ‘intelligent 7) You read a web page of a company. Where is the actual file located, that you agent’, both alone and in combination are reading? with interactive voice interfaces.

Telektronikk 3/4.1997 59 use of the web. They were also issued the • All subjects, also those who had never intelligent agents with the ability to adapt questionnaire, both before and after the used the web, were able to give some to our needs will be our future communi- course. form of explanation about the web. cation intermediaries. Today’s ubiquitous telephone numbers may soon be replaced • Some subjects used the ‘Surf’ or The novice users who had not used the by address systems based on the sub- ‘Print’ metaphors even though they web before as well as the people taking scriber’s name, place of work or other actually had more exact knowledge the WWW course were able to explain easily remembered distinguishing charac- about the browser/web interaction their notions about the functioning of the teristic; the ungainly digit strings re- (Group 3). web with some simple model. These quired to access complex services may models fell into three basic categories, • Some subjects used the ‘Surf’ or soon be superseded by object-oriented which were characterised by different ‘Print’ metaphors, but could not graphic user interfaces; the interactive levels of understanding what was really always predict correctly the behaviour voice-based services requiring key- going on: of the browser (Group 2). presses in response to prompts, will soon give way to intelligent agents that discern 1 Fetch files to my computer-explana- • Training in the WWW course had no the subscriber’s needs and wishes; and tions contained file movement related effect on the explanation model sub- mobile services have already started to terms such as files being copied to own jects used. free subscribers from fixed communica- hard disk. Typical explanations were • The quality of the answers from the tion sites, a development that will only “The browser asks other computers to novice group and the WWW course spread world-wide as low-orbit satellites send files to user’s computer”, “The group were very similar. or stabilised air-ship-borne transponders files shown are copies of the original are introduced. files, copied to the own hard disk”. • Only one of four wrong answers to questions 2–7 by participants of the 2 Surf around-explanations were iden- Internet and the World Wide Web have WWW course were correctly answered tified by usage of own movement already made great inroads into tradi- after the course. related terms that imply that the web tional telecommunication services, and browser temporarily logs onto the other increasingly more of the new services Ranta-aho’s results suggest that a mental computer and reads the web page files. will be based on information techno- model of the web may be created even Typical explanations were “I go to that logies (IT) rather than on the traditional without any practical experience with the page”, “I am in the other computer”. public switched telephone network web. An inaccurate mental model can be (PSTN). This makes it more important 3 Print onto my screen-explanations very resistant to correct information. than ever before to provide simple, easy- fall between the first two and are Those who have a correct mental model to-learn user interfaces that will not characterised by wordings like “from tend not to be able to predict important exclude users, especially people who are the other computer, my browser shows aspects of the web browser behaviour elderly or impaired. Actually, IT-based the file on the screen”. They do not more correctly than those who hold an services are inherently easier to configure suggest “being logged as user” to the incorrect model are. to individual user needs than the tradi- other computer nor do they suggest tional telephone services. “files being transferred to their own In conclusion Ranta-aho stresses the computer”, but just “being shown on importance of the user interface reflect- The cost/benefit of improved usability of its screen”. ing the users’ task instead of the system’s telecommunication services has not been properties [80]. This can lead to interest- extensively documented, except for inter- The interviewees’ replies fell into three ing results if some users generate system nal operator procedures, where substan- groups: architecture models based on task level tial savings have been achieved by rather metaphors. Even if users do not need to 1 Offered ‘Fetch files’-explanations and small measures. The same potential sav- understand technical details of computer correct answers to all seven questions ings are possible by paying attention to networks, they may soon have to under- (2 of the 10 novices, 4 of the 20 the interface for the end-user. This issue stand the basics of computer networking WWW course participants). has now been placed firmly on the to be able to work correctly and safely on agenda, and there is now an increasing 2 Offered the ‘Surf’ or ‘Print’-explana- the web. Such information can also be awareness among operators and de- tions and had one or more wrong presented on the user task level by using velopers of the potential benefits and answers to the seven questions (8 of “I sit at my computer and all information savings that improved usability can give. the 10 novices, 12 of the 20 WWW flows to me”-metaphor instead of “I surf group). around the world”-metaphor. Finally, standardisation has been men- 3 Also gave the ‘Surf’ or ‘Print’-expla- tioned by some of the authors. In the nations, but actually knew the right 9 Conclusions days of the state-owned telephone admin- answers to the questions. When istration monopolies, the end-user had no prompted they explained “he really Several of the new solutions reviewed way of influencing the design of telecom does not go anywhere but his computer above may have significant effects on services or the pricing policies, except fetches the file. But he is used to talk our future use of telecommunications and very indirectly through general elections. about it like this” (0 of the 10 novices, on the way we will interact with the new Standardisation of telecommunications in 4 of the WWW group). systems and services. Gone are soon the those days was carried out in Europe by days of the simple telephone number CEPT (Conférence Européenne des The author also notes that: dialling, and instead voice recognition, Administration des Postes et Télécommu- graphical user interfaces, the Internet and nications) and globally by CCITT

60 Telektronikk 3/4.1997 (Comité Consultatif International Télé- 10 References 9 FIPA. Foundation for Intelligent graphique et Téléphonique, now ITU-T) physical agents” [WWW site] http:// to achieve trans-border and trans-service 1 Nordby, K (ed.). Proceedings of the www.cselt.stet.it/fipa (visited Febru- interoperability, but users and consumers 16th International Symposium on ary 1997). had virtually no way of influencing the Human Factors in Telecommunica- standards adopted. When the open Euro- tions (HFT’97), Oslo, May 12–16, 10 Goldstein, M. A common functional pean telecommunication market was cre- 1997. Oslo, Telenor, 1997. (ISBN interface for all Types of voice ated, the Commission of the European 82-994236-0-0.) response services. In: K. Nordby Communities (CEC) set up the European (ed.). Proceedings of the 16th Inter- Telecommunication Standards Institute 2 ETSI ER 145. Human Factors : national Symposium on Human (ETSI) to create European regional European harmonization of network Factors in Telecommunications telecommunication standards. ETSI, generated tones. Part 1: A review (HFT’97), Oslo, May 12–16, 1997. however, was not only open to the tele- and recommendations. Sophia Oslo, Telenor, 1997, 33–40. (ISBN com operators and the regulatory Antipolis, ETSI, 1997. 82-994236-0-0.) authorities, but was also open to industry, universities, research organisations, and 3 ETSI ER 145. Human Factors : 11 Schmidt, M et al. User perspectives even to private organisations and statute European harmonization of network on the security of access data, Opera- administrations, including user groups generated tones. Part 2: Listing and tor handover procedures and ‘insult and consumers’ organisations. analysis of European, World and rate’ for speaker verification in auto- standarized tones. Sophia Antipolis, mated telephone services. In: K. With the increasing deregulation of the ETSI, 1997. Nordby (ed.). Proceedings of the telecommunications sector and the grow- 16th International Symposium on ing privatisation of the previously state- 4 Gagliardi, D. Report on the audible Human Factors in Telecommunica- controlled operators, the main emphasis tones in the telephone service in the tions (HFT’97), Oslo, May 12–16, of the operators has shifted more towards EC countries. Report from Contract 1997. Oslo, Telenor, 1997, 41–48. making profit than to provide proper ser- 48159, DG XIII. Brussels, Commis- (ISBN 82-994236-0-0.) vices. One way for the operators to min- sion of the European Communities, imise overheads has been to decrease April 1993. 12 Charry, M C, Zuñiga, M C, Mejia, J participation in standardisation activities. J. An experience of a telecom CTI However, new actors are now rapidly 5 ETSI ETR 170. Generic user control system. In: K. Nordby (ed.). Pro- filling this void. Industry has already procedures for telecommunications ceedings of the 16th International been mentioned, as have universities and terminals and services. Sophia Symposium on Human Factors in research groups, but the most recent Antipolis, ETSI, 1995. Telecommunications (HFT’97), Oslo, actors are the user groups and con- May 12–16, 1997. Oslo, Telenor, sumers’ organisations. ETSI has even 6 Anderson, D M, Ferris, A, Mellors, 1997, 49–56. (ISBN 82-994236-0-0.) established its own User Group to super- W. Human factors in ETSI : use of vise standardisation and see that the tones in public telephone networks. 13 Hamnes, K et al. Dialogue design for users’ perspective is maintained in the In: K. Nordby (ed.). Proceedings of numeric paging using automatic new standards. It may turn out to be a the 16th International Symposium on speech recognition. In: K. Nordby short-sighted, ill-advised and rather Human Factors in Telecommunica- (ed.). Proceedings of the 16th Inter- costly decision by the telecom operators tions (HFT’97), Oslo, May 12–16, national Symposium on Human Fac- to reduce participation in standardisation 1997. Oslo, Telenor, 1997, 3–14. tors in Telecommunications and leave the arena open to the special (ISBN 82-994236-0-0.) (HFT’97), Oslo, May 12–16, 1997. interests of user groups and consumer’s Oslo, Telenor, 1997, 57–64. (ISBN organisations. Actually, the interests of 7 Helmreich, R. Ergonomic user inter- 82-994236-0-0.) the telecom operators and the user organ- faces on telephones – are they worth isations may often coincide, but the tech- it? In: K. Nordby (ed.). Proceedings 14 Fay, D. User acceptance of automatic nical solutions that could be proposed by of the 16th International Symposium speech recognition in telephone ser- the user organisations as part of the new on Human Factors in Telecommuni- vices. Proceedings of the Inter- standards may not always be optimal for cations (HFT’97), Oslo, May 12–16, national Conference on Spoken Lan- the efficient and economic provision of 1997. Oslo, Telenor, 1997, 15–20. guage Processing, Yokohama, 1994, telecommunication services. It is there- (ISBN: 82-994236-0-0.) 1303–1306. fore important that the operators find a strategy for participation in standardisa- 8 Dieterich, H, Steiner, D. Intelligent 15 Karis, D. Speech recognition sys- tion work in close co-operation with the agents as appointment managers liv- tems: Performance, preference, and user representatives. ing in public networks. In: K. Nordby design. In: K. Nordby (ed.). Proceed- (ed.). Proceedings of the 16th Inter- ings of the 16th International Sympo- national Symposium on Human Fac- sium on Human Factors in Telecom- tors in Telecommunications munications (HFT’97), Oslo, May (HFT’97), Oslo, May 12–16, 1997. 12–16, 1997. Oslo, Telenor, 1997, Oslo, Telenor, 1997, 23–30. (ISBN 65–72. (ISBN 82-994236-0-0.) 82-994236-0-0.)

Telektronikk 3/4.1997 61 16 Kosko, B. Hidden patterns in com- Oslo, Telenor, 1997, 91–98. (ISBN 34 Contin, L et al. Impact of video cod- bined and adaptive knowledge net- 82-994236-0-0.) ing degradation on emotion recogni- works. International Journal of tion. In: K. Nordby (ed.). Approximate Reasoning, 2, 1988, 26 Mowbray, G H. Choice reaction Proceedings of the 16th International 377–393. times for skilled responses. Quarterly Symposium on Human Factors in Journal of Experimental Psychology, Telecommunications (HFT’97), Oslo, 17 Kosko, B. Neural Networks and 12, 1960, 193–202. May 12–16, 1997. Oslo, Telenor, Fuzzy Systems. New York, Prentice- 1997, 125–131. (ISBN 82-994236-0- Hall, 1995. 27 Goldstein, M. Classification of 0.) methods used for assessment of text- 18 Zajicek, M et al. Answering vague to-speech systems according to the 35 Tetzchner, S von. The use of 64 enquiries in computer aided tele- demands placed on the listener. Kbit/s videophones for sign phony. In: K. Nordby (ed.). Proceed- Speech Communication, 16 (3), language. In: S. von Tetzchner (ed.). ings of the 16th International Sympo- 1995, 225–244. Issues in Telecommunication and sium on Human Factors in Telecom- Disability : COST 219 Project. Brus- munications (HFT’97), Oslo, May 28 Delogu, C, Paoloni, A. Evaluation of sels, Commission of the European 12–16, 1997. Oslo, Telenor, 1997, speech synthesis systems for Communities, 1994, 327–339. (ISBN 75–82. (ISBN 82-994236-0-0.) telecommunication services. In: K. 92-826-3128-1.) Nordby (ed.). Proceedings of the 19 ITU-T. Modulated Noise Reference 16th International Symposium on 36 Kamata, K. Sign language conversa- Unit (MNRU). Recommendation Human Factors in Telecommunica- tion through videophone in ISDN P.81, Blue Book, 5, 1989, 198–203. tions (HFT’97), Oslo, May 12–16, service. Proceedings of the 2nd 1997. Oslo, Telenor, 1997, 99–106. European Conference on the Ad- 20 ITU-T. An On-going Series of Sub- (ISBN 82-994236-0-0.) vancement of Rehabilitation Technol- jective Experiments to Assess Speech ogy (ECART 2), Stockholm, Sweden, Output from Text-to-speech Systems. 29 Walker, H W, Sproull, L, Subramani, 1993. COM-12-18-E, June 1993. R. Using a human face in an interface. In: Proceedings of CHI’94, 37 Kamata, K, Hirama, A. The use of 21 ITU-T. Report of the First Meeting of Human Factors in Computing Sys- 64kbit/s videophones for hearing Study Group 12 (Geneva, 10–19 May tems. Boston, CHI, 1994, 65–91. impaired people. In: Proceedings of 1993), Part B : Draft ITU-T Recom- the 15th International Symposium on mendation P.85 : A Method for Sub- 30 Taylor, I C et al. User trust in talking Human Factors in Telecommunica- jective Performance Assessment of heads as providers of information in tions (HFT’95), Melbourne, Aus- the Quality of Speech Voice Output multimedia interfaces. In: K. Nordby tralia, 1995. Devices. COM-12-R-6-E, June 1993. (ed.). Proceedings of the 16th Inter- national Symposium on Human Fac- 38 Kamata, K, Hirama, A, Oguro, S. 22 Watanabe, T. TFW-PM reference for tors in Telecommunications ISDN videotelephones for the people subjectively assessing quality of syn- (HFT’97), Oslo, May 12–16, 1997. with hearing disabilities. Proceedings thetic speech. In: K. Nordby (ed.). Oslo, Telenor, 1997, 109–116. (ISBN of the 3rd European Conference on Proceedings of the 16th International 82-994236-0-0.) the Advancement of Rehabilitation Symposium on Human Factors in Technology (ECART 3), Lisbon, Por- Telecommunications (HFT’97), Oslo, 31 McGurk, H, MacDonald, J. Hearing tugal, 1995, 276–278. May 12–16, 1997. Oslo, Norwegian lips and seeing voice. Nature, 264, Telecommunications Authority, 1976, 746–748. 39 Kamata, K, Yamashita, M, Hirama, 1997, 83–90 (ISBN 82-994236-0-0.) A. Sign language conversation and 32 Sekiyama, K, Tohkura, Y. McGurk picture quality of 64Kb/s ISDN 23 Hick, W E. On the rate of gain of effect in non-English listeners: few videophones. In: K. Nordby (ed.). information. Quarterly Journal of visual effects for Japanese subjects Proceedings of the 16th International Experimental Psychology, 4, 1952, hearing Japanese syllables of high Symposium on Human Factors in 11–24. auditory intelligibility. Journal of Telecommunications (HFT’97), Oslo, Acoustic Society of America, 90 (4), May 12–16, 1997. Oslo, Telenor, 24 Davis, R, Moray, N, Triesman, A. 1991, Pt 1. 1997, 133–138. (ISBN 82-994236-0- Imitative responses and the rate of 0.) gain of information. Quarterly Jour- 33 Contin, L et al. Methodological nal of Experimental Psychology, 13, studies on subjective evaluation of 40 Fosse, T V, Dolvik, J. Sign language 1961, 78–79. audio-visual quality. In: K. Nordby video for hearing impaired, com- (ed.). Proceedings of the 16th Inter- patible with digital video broadcast- 25 Thomas, B. Choice reaction times in national Symposium on Human ing. In: K. Nordby (ed.). Proceedings voice applications. In: K. Nordby Factors in Telecommunications of the 16th International Symposium (ed.). Proceedings of the 16th Inter- (HFT’97), Oslo, May 12–16, 1997. on Human Factors in Telecommuni- national Symposium on Human Oslo, Telenor, 1997, 117–124. (ISBN cations (HFT’97), Oslo, May 12–16, Factors in Telecommunications 82-994236-0-0.) 1997. Oslo, Telenor, 1997, 165–166. (HFT’97), Oslo, May 12–16, 1997. (ISBN 82-994236-0-0.)

62 Telektronikk 3/4.1997 41 Suwita, A, Mühlbach, L. Video- 49 Zajicek, M, Powell, C. Accessing the Symposium on Human Factors in communications as a medium for World Wide Web by telephone. In: Telecommunications (HFT’97), Oslo, informal communication? In: K. K. Nordby (ed.). Proceedings of the May 12–16, 1997. Oslo, Telenor, Nordby (ed.). Proceedings of the 16th International Symposium on 1997, 217–224. (ISBN: 82-994236- 16th International Symposium on Human Factors in Telecommunica- 0-0.) Human Factors in Telecommunica- tions (HFT’97), Oslo, May 12–16, tions (HFT’97), Oslo, May 12–16, 1997. Oslo, Telenor, 1997, 177–184. 56 Delvert, J. Verification of quality 1997. Oslo, Telenor, 1997, 141–147. (ISBN 82-994236-0-0.) requirements on video-telephony for (ISBN 82-994236-0-0.) sign language : videotelephony and 50 Beattie, W, Hine, N, Arnott, J. Dis- relay service for sign language. A 42 Hopf, K, Runde, D. 3D-desktop-distributed assistive communication de- comparison between 4 projects with plays for multimedia. In: K. Nordby vices for non-speaking users. In: K. different technique, bandwidth and (ed.). Proceedings of the 16th Inter- Nordby (ed.). Proceedings of the services. In: K. Nordby (ed.). Pro- national Symposium on Human Fac- 16th International Symposium on ceedings of the 16th International tors in Telecommunications Human Factors in Telecommunica- Symposium on Human Factors in (HFT’97), Oslo, May 12–16, 1997. tions (HFT’97), Oslo, May 12–16, Telecommunications (HFT’97), Oslo, Oslo, Telenor, 1997, 149–156. (ISBN 1997. Oslo, Telenor, 1997, 185–190. May 12–16, 1997. Oslo, Telenor, 82-994236-0-0.) (ISBN 82-994236-0-0.) 1997, 225–231. (ISBN 82-994236-0- 0.) 43 Prussog, A, Mühlbach, L, Böcker, M. 51 Hine, N, Beattie, W, Arnott, J. Study Telepresence in stereoscopic video- of picture annotation as a means of 57 Sallnäs, E-L, Hedman, L-R, Virta- conferencing. In: Proceedings of the assisting non-speaking people to use nen, M T. Assessment of user needs 14th International Symposium on telecommunications services. In: K. for future work practices. In: K. Human Factors in Telecommunica- Nordby (ed.). Proceedings of the Nordby (ed.). Proceedings of the tions (HFT’93). Darmstadt, R.v. 16th International Symposium on 16th International Symposium on Deckers Verlag, 1993. Human Factors in Telecommunica- Human Factors in Telecommunica- tions (HFT’97), Oslo, May 12–16, tions (HFT’97), Oslo, May 12–16, 44 Runde, D, Suwita, A. Stereoscopic 1997. Oslo, Telenor, 1997, 191–198. 1997. Oslo, Telenor, 1997, 267–274. pointing in desktop applications. In: (ISBN: 82-994236-0-0.) (ISBN 82-994236-0-0.) K. Nordby (ed.). Proceedings of the 16th International Symposium on 52 Hine, N. People with special needs 58 Böhm, A et al. Use of collaboration Human Factors in Telecommunica- and service aspects. In: P. Byerley & tools in distributed research projects tions (HFT’97), Oslo, May 12–16, S. Connell (eds.). Integrated Broad- now and in future. In: K. Nordby 1997. Oslo, Telenor, 1997, 157–164. band Communications : Views from (ed.). Proceedings of the 16th Inter- (ISBN 82-994236-0-0.) Race, Usage Aspects. North-Holland, national Symposium on Human Fac- Elsevie, 1992, 201–230. (ISBN 0- tors in Telecommunications 45 Bias, R G, Mayhew, D J (eds.). 444-89391-1.) (HFT’97), Oslo, May 12–16, 1997. Cost–Justifying Usability. Boston, Oslo, Telenor, 1997, 275–284. (ISBN Academic Press, 1994. ISBN: 0-12- 53 Hine, N, Beattie, W, Arnott, J. Fac- 82-994236-0-0.) 095810-4.) tors affecting usability of communications services by people with dis- 59 Honda, S et al. A virtual office en- 46 Petrie, H, Fabrizi, P, Homatas, G. abilities. In: K. Nordby (ed.). Pro- vironment based on a shared room Report on user requirements for the ceedings of the 16th International realizing awareness space and trans- accessibility of graphical user inter- Symposium on Human Factors in mitting awareness information. In: K. faces by blind people. In: Graphical Telecommunications (HFT’97), Oslo, Nordby (ed.). Proceedings of the User Interfaces for Blind Persons : May 12–16, 1997. Oslo, Telenor, 16th International Symposium on TIDE Pilot Action Project 103. 1997, 201–208. (ISBN 82-994236-0- Human Factors in Telecommunica- Report to the CEC. London, Royal 0.) tions (HFT’97), Oslo, May 12–16, National Institute for the Blind, 1993. 1997. Oslo, Telenor, 1997, 287–295. 54 Van Mele, I et al. Mobile access to (ISBN 82-994236-0-0.) 47 Stephanidid, C, Savidis, A, Acoumi- WWW services for university stu- anakis, D. Tools for user interfaces dents who are motor impaired : a 60 Nilles, J. Making Telecommuting for all. Proceedings of Second TIDE qualitative study on the man/machine Happen : a guide for telemanagers Congress. Amsterdam, IOS Press, interface. In: K. Nordby (ed.). Pro- and telecommuters. International 1995. ceedings of the 16th International Thomson Publishing, VNR Com- Symposium on Human Factors in puter Library, 1994. 48 Laux, L et al. Designing the World Telecommunications (HFT’97), Oslo, Wide Web for people with disabili- May 12–16, 1997. Oslo, Telenor, 1997, 61 Nilles, J. Telecommuting and traffic ties : A user centred design approach. 209–216. (ISBN 82-994236-0-0.) reduction. ECTF News, #6, 1994. Proceedings of ASSETS’96, Second Annual ACM Conference on Assistive 55 Hellström, G. Quality measurement 62 Matarazzo, G. Telework in the Technologies. New York, ACM, on video communication for sign lan- Municipality of Rome: a case study. 1996. guage. In: K. Nordby (ed.). Pro- In: K. Nordby (ed.). Proceedings of ceedings of the 16th International the 16th International Symposium on

Telektronikk 3/4.1997 63 Human Factors in Telecommunica- 68 Heim, A W. AH Series, NFER-NEL- 16th International Symposium on tions (HFT’97), Oslo, May 12–16, SON, 1970. Human Factors in Telecommunica- 1997. Oslo, Telenor, 1997, 297–303. tions (HFT’97), Oslo, May 12–16, (ISBN: 82-994236-0-0.) 69 Gronwall, D. Paced Auditory Serial 1997. Oslo, Telenor, 1997, 401–408. Addition Task : A measure of re- (ISBN 82-994236-0-0.) 63 Papa, F, Spedaletti, S. New broad- covery from concussion. Perceptual band technologies in tele-education: and Motor Skills, 44, 1977, 367–373. 75 Svendsen, G et al. A method for cre- A human factors laboratory experi- ating and introducing new telecom ment. In: K. Nordby (ed.). Proceed- 70 Lezack, M D. Neuropsychological services and products. In: K. Nordby ings of the 16th International Sympo- Assessment, 2nd edition. Oxford, (ed.). Proceedings of the 16th Inter- sium on Human Factors in Telecom- Oxford University Press, 1983, national Symposium on Human Fac- munications (HFT’97), Oslo, May 422–429. tors in Telecommunications 12–16, 1997. Oslo, Telenor, 1997, (HFT’97), Oslo, May 12–16, 1997. 305–312. (ISBN 82-994236-0-0.) 71 Gustafsson, N-E. Intranetworking Oslo, Telenor, 1997, 409–413. (ISBN with users – Fast & facile feedback. 82-994236-0-0.) 64 Fleet, G, Talbott, D, Winlow, T. In: K. Nordby (ed.). Proceedings of Graphical user interface design for the 16th International Symposium on 76 Lindgaard, G. Wow : 568 smart fea- the user delivering value to the cus- Human Factors in Telecommunica- tures on your PABX : What really tomer. In: K. Nordby (ed.). Proceed- tions (HFT’97), Oslo, May 12–16, determines the uptake of technology? ings of the 16th International Sympo- 1997. Oslo, Telenor, 1997, 331–337. In: Proceedings of the 14th Interna- sium on Human Factors in Telecom- (ISBN 82-994236-0-0.) tional Symposium on Human Factors munications (HFT’97), Oslo, May in Telecommunications (HFT’95). 12–16, 1997. Oslo, Telenor, 1997, 72 Ling, R, Hareland, M. The cost of Darmstadt, R.v. Deckers Verlag, 315–322. (ISBN 82-994236-0-0.) being social: User expectations of 1993. metering unit displays and cordless 65 Love, S, Foster, J C, Jack, M A. telephones. In: K. Nordby (ed.). Pro- 77 Herstad, J, Hüttenrauch, H, von Assaying and isolating individual dif- ceedings of the 16th International Niman, B. Now 569 Smart features ferences in automated telephone ser- Symposium on Human Factors in on your PBX: Increasing usability vices. In: K. Nordby (ed.). Proceed- Telecommunications (HFT’97), Oslo, through Computer Telephony Inte- ings of the 16th International Sympo- May 12–16, 1997. Oslo, Telenor, gration …? In: K. Nordby (ed.). Pro- sium on Human Factors in Telecom- 1997, 365–372. (ISBN: 82-994236- ceedings of the 16th International munications (HFT’97), Oslo, May 0-0.) Symposium on Human Factors in 12–16, 1997. Oslo, Telenor, 1997, Telecommunications (HFT’97), Oslo, 323–330. (ISBN 82-994236-0-0.) 73 Schyguda, G. Societal trends and May 12–16, 1997. Oslo, Telenor, their impact on telecommunication 1997, 427–434. (ISBN: 82-994236- 66 Egan, D E, Gomez, L M. Assaying, markets. In: K. Nordby (ed.). Pro- 0-0.) isolating and accommodating indi- ceedings of the 16th International vidual differences in learning a com- Symposium on Human Factors in 78 Mané, A, Rabin, M. Meet your complex skill. In: Individual Differences Telecommunications (HFT’97), Oslo, munication intelligent agent on the in Cognition, 2, Academic Press, May 12–16, 1997. Oslo, Telenor, World Wide Web. In: K. Nordby 1985. 1997, 383–390. (ISBN 82-994236-0- (ed.). Proceedings of the 16th Inter- 0.) national Symposium on Human 67 McCrae, R R, Costa, P T. Validation Factors in Telecommunications of the Five-Factor Model of Person- 74 Ranta-aho, M, Leppinen, A. Match- (HFT’97), Oslo, May 12–16, 1997. ality across instruments and ob- ing telecommunication services with Oslo, Telenor, 1997, 435–441. (ISBN servers. Journal of Personality and user communication needs. In: K. 82-994236-0-0.) Social Psychology, 52, 1987, 81–90. Nordby (ed.). Proceedings of the 79 Ranta-aho, M. WWW and the surfing metaphor: harmful for the novice user? In: K. Nordby (ed.). Proceed- ings of the 16th International Sympo- sium on Human Factors in Telecom- munications (HFT’97), Oslo, May 12–16, 1997. Oslo, Telenor, 1997, Knut Nordby is Senior Research Psychologist at 443–450. (ISBN: 82-994236-0-0.) Telenor R&D, Kjeller, where he has worked since 1985. He is Chairman of the ETSI Technical Committee Human Factors (TC HF), of which he was a founding 80 Nielsen, J. Usability engineering. NJ, member. He is member of the Permanent Steering Academic Press, 1993. Committee of Human Factors in Telecommunications (HFT) and Adjunct Professor teaching Man-Machine Interaction at UNIK, Center for Technology at Kjeller, University of Oslo. e-mail: [email protected]

64 Telektronikk 3/4.1997 Feedback tones in public telephone networks: Human Factors work at ETSI DONALD M. ANDERSON

A problem for users? luck for the majority of users. An example of early concern All telephone users are familiar with for the user, from nearly 30 STATE - Waiting for user´s control action feedback from the network – they lift the years ago is given by Karlin handset, hear what they know as a [1], when the CCITT discussed ‘dialling’ tone, dial, then hear a ‘ringing’ a contribution based on experi- CONTROL ACTION tone. Or do they? Is it a ringing tone or a mental evidence from five dif- ‘congestion’ tone, or something else they ferent countries, which showed INDICATE - Feedback on control and system changes do not recognise? The effect may be the the degree of confusion that same, if they are not connected with then existed between identifi- speech, they will probably hang up and cation of ‘ringing’ and ‘busy’ INDICATE - Prompt on next expected inputs try later. Users mostly do not understand telephone signals. the meaning of all the tones they may hear. This problem is much worse when STATE - Waiting for next user´s control action The ITU-T (formerly CCITT), dialling to another country or when visit- was from early on aware of the ing another country and trying to call potential problem that diffe- home. Even worse, in a payphone they Figure 1 Simple model of user procedure rences between national net- may lose any money they may have work tones would cause for inserted, because of such misunder- users, and by promulgating standing. ‘Recommendations’, tried to standardise the parameters for audible tones for use A previous generation of users simply in public networks, both by describing lifted the phone receiver off the hook and the function represented by the tone code, a human operator asked you for the such as ‘dial’, ‘ring’, or ‘busy’, and also number you wish to call, perhaps the aspects such as audio frequency, cadence The complete activity undertaken by the ultimate form of feedback. The operator and other signal parameters to ensure sig- team appears in a two-part Technical would even tell you that the called party nal compatibility between networks, but Report from ETSI ([3] and [4]). The sec- was busy or not replying. Then around not necessarily user understanding. ond part contains what is probably the the 1920s, automation allowed you to most comprehensive analysis and up-to- dial your own number, so feedback by date listing of tones used in the EU, in The EU in 1993 recognised that the pro- tones was introduced. Such a simple user the countries of the ETSI members and in liferation of tones caused by continuing procedure may be represented in a classic other countries world-wide published to expansion of telephone networks and ser- ‘indicate-control-indicate’ model, as date. The analysis was of special interest vices world-wide had led to potential for illustrated in Figure 1. because it revealed that coding was confusion and error by the users and in a mainly by frequency and especially by review carried out by Gagliardi [2] he cadence. Two examples of tables from Under normal conditions, a user no proposed that standardisation and har- the analysis of European tones are given longer had any need for operator inter- monisation of network tones between EU in Annex 1. vention, except for ‘trunk calls’. This and European countries should take was good human factors practice applied place. This paper describes the aims and Six tones were defined in the Mandate long before the expression became recog- methods used to review the problem and and specifically suggested as priority nised, but at that time, only a limited arrive at possible proposals for change. candidates for standardisation, including: number of tone codes was necessary. The The work was undertaken by a team of situation worsened when subscribers human factors specialists working to an • dial, ringing, busy, special informa- became able to dial their own calls into EU Mandate, under contract to the Euro- tion, call waiting and pay tones. networks overseas, who had adopted pean Telecommunications Standards different ‘foreign’ tones. As services Institute (ETSI). A further set of four tones were proliferated and became more automated suggested as possible candidates, for direct access by a subscriber utilising Review of the problems including: a telephone dial, so even more informa- • special dial tone, positive indication tion was, of necessity, provided in the A number of specific tasks was proposed tone, intrusion tone and congestion network for presentation to the user as in the EU Mandate. These included: tone. tones. • a review of the research and reports of the Human Factors of service tones Today, in some 15 countries in the EU, Additional tones for other functions were and 174 world-wide, some 228 different • a review of the standards originating also to be considered for standardisation tones are defined as in use in Europe from the ITU-T, CEPT and ETSI if they were regarded as necessary, or alone. In addition to feedback to the others may be excluded. Technical and • a review of the technical aspects of human user, audible tones may also be economic factors were also to be taken tone production in networks, and used by connected terminals for auto- into account together with a limited matic recognition. For humans, recog- • a review of actual national regulations evaluation for cost and feasibility and nition of the meaning of some tones is in force in the countries of the Euro- time scale for technical implementation today largely a matter of guesswork and pean Union. of possible changes.

Telektronikk 3/4.1997 65 Ideally, Gagliardi’s contentions should Talk Phase have been subjected to user testing to find out the degree of confusion or otherwise existing between tones and between user groups as he defined them (he de- Interaction fined two groups of users, those at home with System phoning to a different country and those abroad phoning home). This would have established a degree of usability which Progress Failure to Prompt for could have been compared with a normal human factors criterion of requiring Progress Action 95 percent of the target population being Basic Basic able to discriminate accurately between the selected tones on 95 percent of occasions. The tests would have to include Sophisticated Sophisticated representative samples of the cultural mix in EU countries, age, sex and relevant telephone experience; a task of daunting proportions. Progress Failure to No Action Progress In human factors and practical terms, Basic Basic Required however, the only basis for criteria against which to test the tones were the data extracted from the reviews. During a brainstorming session by the study team Sophisticated Sophisticated many individual criteria were examined one at a time, such as the effect of tone characteristics like frequency and cadence, and the understanding of the Figure 2 User goal based model meanings of tones. Attempts to combine this data in some way led to the concept of user modelling. From consideration of the sequence of tasks performed when making a telephone call, the first model was based on users’ goals, and is illu- The Mandate, whilst giving broad guid- be employed in any comparison process strated in Figure 2. ance in terms of reviews that should be to evaluate the target tones in terms of carried out, gave no particular hint as to possible need for harmonization. how this mass of review material should However, this neglected the requirements of the telephone network, so that a task analysis was performed and consideration given to the indicate-control-indicate model already postulated for telephone user interaction. This gave rise to a series of models, which culminated in a final so-called human factors model, where not only were users alternative actions included, but also the relevant phases of a telephone call from the network point "Good" Prompt "BAD" of view, as shown in Figure 3; for a better Feedback Feedback understanding of the content of each box, Can Must the model development is described later. When the human factors data about user Reassuring Worrying perception of tones was also included, it WAIT SIGNAL ABORT Calm Disturbing seemed that a definitive model had emerged Low freq. High freq. Not urgent Urgent against which tones could be tested. Slow cadence Fast cadence Smooth Rough Long period Short period Pleasant Unpleasant Residual doubts about the validity of the Continuing model remained, however, that it was not TALK possible to satisfy without data from user Low duty cycle testing, so that it was not appropriate to use the model alone as the basis for deciding on harmonization issues. Whilst it Figure 3 Tone characteristics model used in final audit

66 Telektronikk 3/4.1997 was not feasible to use the model as the Successful telecommunications depends may suggest differing characteristics for sole criterion, it was possible to use the on the quality of the user/terminal/net- tones. For example, a crude distinction data from the reviews as well, in a linear work interaction necessary for the pur- could be developed where feedback tones comparison process termed an audit. In poses of establishing a call, modifying it are slow and reassuring, but prompt tones this way almost all items relevant to (by invoking a supplementary service), are fast and alerting. harmonization, not only human factors, or clearing it down. The quality of this would be given due consideration and interaction may be judged from the view Other indications are not for feedback on form the basis for rational recommenda- of the network (or rather its designer) on a user’s actions or to prompt a user to tions. the basis of whether the user responds to make a control input, but to prompt the indications as desired, and from the view user that they may have to change their of the user on the basis of comprehensi- behaviour to accommodate the changed/ User perception and bility of such indications and the ease of unexpected status of the current connec- use. A good system should meet the description of tones tion, e.g. Conference tone when a new desires of both parties. party joins a conference. A review of the human factors literature covered both general research and reference sources more specifically User-network interaction, In ordinary telephony, the number of related to telecommunications. The task analysis and feedback options open to the user for controlling a findings were grouped into a number of call is limited and may be summarised as areas as they relate to different aspects of the SWAT model: The procedural model in Figure 1 was the perception, discrimination and recog- based on a task analysis of the user-tele- • Signal – to the terminal/network or ser- nition of information tones. The review phone network interaction, as a paradigm vice: includes going off-hook, dialling included some work on music and audi- for user procedures. Each indication numbers, pressing function keys, etc. tory icons, for instance, which allowed a should consist of two components, an broader view to be taken of harmonisa- • Wait – until the telephone, network, ser- indication of the underlying system status tion possibilities and options for change. vice or other user/s has done something and an indication of the status of the required control. • Abort – return the terminal, network The review first established the basic and/or service to idle facts about human perception of sounds. When this principle is integrated into a • Talk – all telecommunication includ- As being highly relevant in relation to procedural task involving the system ing fax and data. tones in the telephone network, limits of (terminal/network) progressing through human capacity for coding and tone dis- a number of intermediary states in The feedback indications used to inform crimination were identified, as determi- response to user control actions, the and prompt the user need to differentiate nants of differentiation between tones required indications become chained between which of these base level and the limits to confusion. Considera- together. Any single indication given options is appropriate, and possibly also tions of recognition time and propensity after a user’s control action can thus have which is the most appropriate choice of to making errors were taken into account more than one function: it can be both the various signals for the user to send. as being relevant to the efficiency of the feedback on the changes to the system/ telephone network, and work on warn- control status and a prompt for the next ings and urgency in similar context. The These complex interactions between expected actions. This distinction is im- question of relating particular tones or users and networks were fully explored portant when considering the roles that their pitch to somewhat abstract concepts early in the project. They were more tones play within a user/terminal/network like prompts or acknowledgement was complex than first thought and led to a interaction, because the different roles particularly relevant, but apparently a succession of models. neglected area of research.

Two issues remained somewhat grey areas from the published data which are important in considering the validity of the model. One is the strength of association of certain tones with abstract concepts like reassuring or disturbing (see Table 1 Task Analysis of existing and proposed tones Figure 3 and later). This question is being studied currently at the Centre for Tone Type When User Choices Communication Interface Research (CCIR) in Edinburgh. CCIR are also performing other work with relevance to Busy Prompt Set-up Hang up or invoke SS - CCBS, UUS tones in exploring preferences for certain types of sound to represent categories of Call waiting Prompt Talk Ignore or Hang up to get new call or information, e.g. simple tones or rising or Attempt call control falling chimes. Another issue is the range of frequencies associated with qualitative Dial Prompt Set-up Send or Dial address expressions like high or low (frequency).

Telektronikk 3/4.1997 67 Meaning of tones and feedback are usually prompts for action; in some Dimension Four – Basic tone / sophisti- cases tones have a dual purpose of both cated tone: Dial tone can be considered Work in the area of audible signals feedback and prompt. A further point to to be a basic tone, giving an indication of suggests that specific characteristics of be considered is whether the tones are reaching the desired system status and sounds impart specific meaning to the applied during the setting up of a call or also prompting for a number input. listener. For instance Pollack [5] has service (when interworking with the net- Special dial tone would be classified as shown that tones with long cadences work) or are applied later during the con- a sophisticated tone giving additional infer ‘goodness’ whereas tones with short versational phase of the call (talk or information on the system status to an cadences infer ‘badness’. In defining telecommunicate). Tones occurring dur- expert user. what attributes should be inherent in a set ing the conversational phase of a call are of telecommunications network tones it generally unexpected and indicate a The benefits of such a model/taxonomy is necessary to consider what information change in the network status which may are as follows: the tones need to impart. or may not be desired by the user. 1 If it is possible to associate each of the dimensions with a unique tone charac- Considering that only four actions in A task analysis of common user inter- teristic, then the specification of a new total are available to the user (Signal, actions with the telecommunications net- tone (or tone set) will be facilitated. Wait, Abort and Talk) and only three work suggested that the feedback given during the call set-up phase, it is tempt- 2 Categorisation as above indicates by tones needs to indicate success or ing to consider whether more than three groupings and/or individual tones failure in achieving the user’s desired tones are required. This idea leads to the which may be considered in more goal; such tones either provide a prompt possibility of a taxonomy of tones into detail for combination with others, or to stimulate an action from the user, or basic groups or families. Within any deletion. indicate that no further action is required. group of tones there can be simple, basic The information given by the tone needs 3 If the tone characteristics define tones (reflecting the less informed user’s to be sufficient to elicit the appropriate whether action is required and whether requirement for simple prompts) and action from the user. Table 1 gives an progress is being made towards the sophisticated tones containing additional example for three of the target tones. goal, then in the worst case situation information permitting the knowledgeable the user may still achieve their goal user to adjust his actions to the prompt. without recognising ANY INDI- Tones occurring at the end of a success- VIDUAL tones, only their characte- ful interaction can be considered to be User goal based model ristics. feedback of the resultant network status, often requiring no action; tones occurring The multi-dimensional user goal based during a series of multiple interactions model in Figure 1 attempted to define the Categorisation and complete ‘set’ of network tones in terms characteristics of tones of the following dimensions: The next step was to consider only the Dimension One – Action first two dimensions of the model and Required / Not Required: On assign to them the tones defined in the hearing the tone in question the Mandate (and some others), with the user is prompted either to carry tones shown in bold being those defined Progress Failure to Progress out a specific action in order to as priority. Prompt for Action Prompt for Action move closer to achieving their goal (which could be setting up Figure 4 shows the assignments which a conversation, setting up a net- were reasonably straightforward, and gave Dial Busy work service, etc.) or to remain some confidence in the applicability of the Call waiting Paytone passive, awaiting the next phase model. The model gave somewhat inde- in the process. Special Dial Special information terminate results for only one tone. Spe- Personal User Identity Congestion cial information tone as defined by ITU-T Negative information Dimension Two – Progress / [6] is used to indicate call failure. In some Failure to progress: On hearing cases it stands on its own and requires that the tone in question, the user is the caller acts to release the call. In other Progress Failure to Progress aware whether or not they have cases it is followed by a message and Action not required Action not required moved closer to achieving their requires the caller to wait for the message goal. before deciding on a course of action. It is Ring Special information thus context dependent and is italicised to show that it does not fall firmly into a sin- Positive information Intrusion Dimension Three – Interaction gle classification. Pre empt with the system / Talk phase: Tones occurring in the interaction phase are normally ex- When attempting to apply the model it pected feedback tones. Tones was necessary to postulate tone charac- occurring during the Talk phase teristics appropriate to the various dimen- are generally unexpected. sions of the model. The following were Figure 4 Tone assignments for dimensions 1 and 2

68 Telektronikk 3/4.1997 identified as possibilities for mapping Table 2 Assignment of tone characteristics from the literature onto the tone model defined above: Frequency: High/low, single/multiple, Progress Failure to progress simultaneous/sequential, rising/falling, discrete steps/ Action required Progress – choose from: Failure to progress – choose from: continuous. Low frequency, Harmonious, High frequency, Discordant Slow cadence, Rising frequency Rapid cadence, Falling frequency Complexity: Harmonic content. Cadence: Fast/slow, Duration, Duty Action required – choose from Action required – choose from Cycle, Continuing, Once/ High Frequency, Discordant, High Frequency, Discordant, twice. Rapid cadence, Continuing Rapid cadence, Continuing

Intensity: Comparative. Action not required Progress – choose from: Failure to progress – choose from: Low frequency, Harmonious, High frequency, Discordant, Attempting to define which dimensions Slow cadence, Rising Frequency Rapid cadence, Falling frequency to apply to the “progress” axis and which to apply to the “action” axis gave rise to Action not required – choose from Action not required – choose from the following interim proposals: Low Frequency, Harmonious, Low Frequency, Harmonious, Slow cadence Slow cadence In the axis ‘progress/failure to progress’;

Progress: • An indication of satisfactory progress should be a ‘good’ tone with reassuring characteristics having a calming effect. • Low frequency, harmonious, slow cadence, Pollack [5], Pollack and Ficks italicised), shown in Table 2. This then information on system status, Prompt [7], Heberle [8], rising frequency? shows that two cells are more or less ‘Can’ suggests some choice for the user concordant but two are also contradic- and Prompt ‘Must’ demands action from tory. This could suggest that this level of the user in order to prevent failure of the Failure to progress: coding discrimination was too simplistic, call.) This version was further simplified, • An indication of failure or imminent that the context in which the published rearranged and populated with tone failure to progress should be a ‘bad’ information was derived was not rele- characteristics derived from the human tone with disturbing characteristics vant, or it could illustrate a conflict be- factors literature, and the subjective de- having a stimulating effect. tween the needs of the system designer scriptions associated with them for each • High frequency, discordant, rapid and those of the user. phase according to the SWAT model, cadence [5,7,8]. and this led to the model used during audit, shown in Figure 3. A number of alternative models were In the axis ‘Action required/action not subsequently investigated, and a number required’; of unsuccessful attempts were made to The final model encompasses virtually select a subset of tone characteristics that all of the dimensions analysed as charac- Action required would satisfy the model in all aspects, but teristics of the user-telephone interaction. • A tone indicating action required it was finally concluded that as the net- Some assumptions relate to user goals in should awaken the user, convey work was not itself user goal related, the this interaction, and others to system urgency and be persistent so as to model in all its variants investigated did goals, because a single model cannot rep- invite action not fully allow mapping of all possible resent them all. Tones assigned to the • High frequency, discordant, rapid goals, or adequately take into account the model appear to be in their right places. cadence, Patterson & Milroy [9], con- limitations on the prompts and informa- If the characteristics of known tones as tinuing (until action performed)? tion to be conveyed to the user. assigned to the model in Figure 2 are examined, they are found largely to conform with the literature-derived figures in Action not required Another model was designed to allow most cases. One exception is Dial Tone, • A tone indicating action not required classification of tones by system state, whose functional definition does not should lull the user into inaction their characteristics, and presentation to make it completely clear whether it the user. It used the original concept of • Low frequency, harmonious, slow, should be considered as a prompt or a OK / Not OK, gave a general indication Patterson & Milroy [9], cadence? feedback tone. Many users merely accept to the user of call progress towards an dial tone as feedback indicating that the (assumed) goal, indicated the phase of Characteristics for these tones from the network (or terminal) is working, or it the call with the addition of talk/non-talk literature were then placed into the four may also be interpreted to suggest that and made a distinction between Feed- cells relevant to the first two dimensions the tone should be within the family back, Prompt ‘Can’ and Prompt ‘Must’. of the model (the tones not allocated ‘Prompt Can’. (Feedback is the simple provision of from experimental evidence being

Telektronikk 3/4.1997 69 Table 3 Possible values for tone characteristics 3 Step 3 then defines the characteristics of the tone postulated by the HF model. Frequency Rhythm Period Other 4 Step 4 identifies the range of charac- ‘Good’ Low < 500 Hz Slow – 1/s Long – 3 s teristics reported for the target tone by the chosen group, i.e. Europe. These ‘Bad’ High > 500 Hz Fast – 2/s Short – 1 s Continuing data may be inspected for coherence, Non-intrusive Talk Phase Intrusive spread of certain characteristics, and any significant modal value. Duty cycle < 5 % Duty cycle > 15 % 5 Step 5 identifies a modal set of characteristics for the tone. By inspection, it may be possible to identify a ‘mode’ group within any one tone – that is, the tone characteristics which the largest number of countries report using. Such a mode may be extracted Nevertheless, there is some confidence in new tone for a given function in future, from any set of tones grouped by using the model to assess the conform- provided its functional description can be economic, regional or geographical ance of each tone to the model, and to matched to a cell in the model. This boundaries. use it to identify the characteristics of a matching is part of the audit procedure new tone for a given function in future, developed and described to examine each 6 Step 6 inspects the existing standards provided its functional description can be of Gagliardi’s six priority tones, in order for that tone, both for network and matched to a cell in the model, as dis- to sensibly determine how they might be terminal generation. cussed in the audit later. ‘harmonised’ in the future. 7 The relevant human factors issues on User requirements For the purposes of this audit, values the users’ expectations, the potential needed to be assigned for the technical for confusion and any usability issues In addition to the use of modelling to characteristics described in the model. are reviewed next. The tone character- identify potential attributes of tones rele- Table 3 gives a preliminary assessment istics are compared for possible con- vant to the type of feedback required, the of these values (it is hoped that they can fusability with other tones in the same reviews carried out early in the study be conformed at some later stage by suit- group or any other group, and usabil- identified some significant user require- able experiment). ity in relation to the task the user is ments, based on specific human user carrying out. For the purposes of the characteristic, such as the auditory sense, audit, only an expert estimation can or memory ability for procedures and Consideration of Gagliardi be made of usability in terms of likely meanings of information tones, or gen- tones for harmonisation confusion with other examples in the eral requirements for the usability of the tone group for users in Gagliardi’s network at a national or international The final stage of the team’s work was to groups A and B (those visitors to a level. Some requirements were identified assess each target tone described in the foreign country dialling internally and in the first place by Gagliardi, others dur- mandate in order to recommend possible users dialling into another country). ing the derivation of the model, and oth- ways of harmonising them in the EU Doubtful cases can only be settled by ers from the literature or from good environment. Data available at this stage usability testing. human factors practice. Some of these included results of the reviews and the user requirements were used as general analysis of European tones, as well as 8 Five options for harmonisation or criteria during the comparison process, or potential for guidance from the final change, labelled a–e, are now audit, described below. models. This assessment was carried out examined in some detail: by means of an audit of each tone, fol- Option a: Use the HF model. In this lowing a detailed procedure consisting of case harmonisation could be achieved Use of the model ten steps: by countries aligning the characteris- The final model encompasses virtually 1 The target tone is first named and tics of their tone with those proposed all of the dimensions analysed as charac- defined by its functional description, by the model. In the case of a gross teristics of the user-telephone interaction. i.e. what it is intended to convey to misfit, the audit would try to de- Some assumptions relate to user goals in the user about a call. At this stage it is termine whether any change could be this interaction, and others to system possible to identify some tones where justified in terms of improvement in goals, because a single model cannot rep- it is not clear either what their func- usability versus cost. resent them all. Tones assigned to the tion is, or that they are using the Option b: Use the existing standards. model appear to be in their right places. wrong description or name. In this case, harmonisation could be achieved by countries changing some 2 The next step is to assign the tone to a There is therefore some confidence in aspects of a tone’s technical charac- position in the HF model, based on its using the model to assess the conform- teristics to align it with a specific functional description. ance of each tone to the model, and to international standard. use it to identify the characteristics of a

70 Telektronikk 3/4.1997 Option c: Use the European mode. Conclusions For ringing tone proposals included As explained above, harmonisation harmonization of cadence and frequency, could be achieved by countries There were two satisfactory outcomes and that requirements for a number of changing harmonizing their tone’s from the study. One was the development possible special ringing tones be de- characteristics to those most com- of a model, based mainly on empirical veloped, such as for special services, monly occurring in Europe (or in findings and a task analysis of the users’ videotelephone call, voice mail, etc. any other grouping). activities in making a telephone call. Applying some research evidence and Option d: Do something different. At the same time, options should be pro- testing known tones to the model appears In an exceptional or new case, it may vided for new dial and ringing tones, to confirm its validity, but some further be desirable to create a tone with based on the model, to assist the provi- testing should be done, particularly with characteristics completely outside an sion of corporate identity in the tone. respect to any possible predictive value existing range, in order to make it the model may have. truly distinguishable, or a characte- It was suggested that benefits might ristic ‘earcon’ may be created. come from introducing a new harmon- Secondly, the many detailed analyses Option e: Do nothing. This would ized high frequency busy tone in prefer- done, and the data extracted from the imply that the tone was usable, would ence to existing usage. Similarly, for spe- human factors literature about human not confuse users in Gagliardi’s Cate- cial information tone, a new tone might perception of tones, lay the foundation gories A or B, and was conformant be developed in line with the HF model for an audit process as a suitable logical with known definitions and standards. in the call failure group to indicate num- basis for recommendations for harmo- In other words, the tone was in effect ber unobtainable or invalid service. nization of network tones. In particular, already harmonised. the systematic consideration of human No other Options are considered, factors issues, such as usability assess- It must be emphasized that the recom- there are other options for change ment, would ensure that if the processes mendations from ETSI to the European which could have been examined, for were to be used for new tones, they Commission remain just that at the time example: would not only fit the model, but be of publication, and no action has been acceptable and useful to representative proposed or discussed. • Different calculations of the Euro- users. pean mode – e.g. by weighting countries by reference to their num- Acknowledgement ber of subscribers, or their volume The analysis of the tone data showed the of traffic, or by their ETSI voting strong dependence at present on using The authors are grateful to the Director- rights. different simple cadences to differentiate General of ETSI for permission to pub- between tones, while neglecting greater • The consideration of a World Mode, lish this paper which is based on work use of frequency. At the same time a however calculated. performed on behalf of ETSI under CEC wide interpretation of the functional Mandate BC-T-308. • The specific use of new tones or definitions of some tones has lead to the earcons. sometimes inappropriate applications in networks, to a degree likely to cause con- However, these five (a–e above) repre- References fusion. The study thus confirmed partly sent the key choices that may be made. the concern by Gagliardi with possible 1 Karlin, J. Summary of results from confusion and error when using the tele- laboratory studies of audible tone 9 The penultimate step reviews the impli- phone within the EU countries. The rec- signals, including data from Ger- cations for each option (a–e), including ommendations, containing mostly minor many, Japan, The Netherlands, crude estimates of cost and benefits and changes to achieve harmonization are set United Kingdom, and United States. the ease with which a possible migra- out in full in the ETSI Technical Report. CCITT WP XIII/1, Temporary Docu- tion path may be established as part of ment No. 6-E. Tokyo, July 1967. an implementation plan. The recommendations particularly 2 Gagliardi, D. Report on the audible 10 The final step derives a set of recom- emphasize the need to harmonize the tones in the telephone service in the mendations in priority order with functional definition of some tones to EC countries. Report from contract respect to the harmonisation of each ensure their correct application and 48159, DGXIII, European Commis- tone. understanding by users and for the ces- sion. Brussels, April 1993. sation of incorrect usage. It is seen as especially important that the same or 3 ETSI TR 101 041-1 ed. 1. Human An example of one such audit, of ringing similar tones are not used within a net- Factors (HF) : European harmoniza- tone is given in Annex 1. work to signify different meanings or tion of network generated tones : services. Part 1 : A review and recommendations, Sophia Antipolis, 1997. In the case of dial tone, it was proposed that a consistent use of a continuous tone 4 ETSI TR 101 041-2 ed. 1. Human of single or dual frequencies be adopted, Factors (HF) : European harmoniza- and gradual change to a harmonized con- tion of network generated tones : tinuous tone of 425 Hz. Part 2 : Listing and analysis of Euro-

Telektronikk 3/4.1997 71 pean, World and Standardized tones. Sophia Antipolis, 1997.

5 Pollack, I. The information of elementary auditory displays. Journal of the Acoustical Society of America, 24, November 1952.

6 ITU-T Recommendation E.180. Technical characteristics of tones for the telephone service. Geneva, 1988.

7 Pollack, I, Ficks, L. The information of elementary multi-dimensional auditory displays. Journal of the Acoustical Society of America, 26, 1954, 155–158.

8 Heberle, W. Human aspects of audible tones for controlling telephone user’s behaviour. Proceedings of 6th International Symposium on Human Factors in Telecommunica- tions, Stockholm, 1972.

9 Patterson, R D, Milroy, R. Auditory warnings on civil aircraft : The learning and retention of warnings. Cambridge, UK Medical Research Council, Applied Psychology Unit, 1980.

Donald M. Anderson is currently working as Research Ergonomist with Rikshospitalet (National Hospital), Oslo, and as the Principal Consultant, Man Machine Technology AS. He has 45 years experience in ergonomics and his interests include standards, mar- itime ergonomics, product design and development and aids for the handicapped. e-mail: [email protected]

72 Telektronikk 3/4.1997 Annex 1 Examples from analysis of European tones

1 Dial tone Functional definition related information. (This definition is consistent with ITU-T The auditory indication to be presented to a user to indicate that Recommendation E.182.) a network connection is available and ready to receive call information and inviting the user to start sending call or service

Table of European Dial Tones

Cadence Period Frequency Country

None reported Malta 1 0,33 – 0,33 – 0,66 – 0,66 1,98 425 Czech Republic, Slovakia 2 0,2 – 0,2 – 0,6 – 1,0 2 425 Italy 3 0,7 – 0,8 – 0,2 – 0,3 2 425 Slovenia, Yugoslavia 4 0,2 – 0,3 – 0,7 – 0,8 2 425 Croatia 5 0,2 – 0,3 – 0,7 – 0,8 2 425 or 450 Greece 1 (ITU) 6 0,2 – 0,8 – 0,7 – 0,3 2 425 or 450 Greece 2 (300 085) 7 0,25 – 0,3 – 0,7 – 0,8 2,05 425 Bulgaria 1 8 0,25 – 0,75 – 0,75 – 1,0 2,75 425 Bulgaria 2 9 Continuous 350+400 Gibraltar 10 Continuous 350+440 United Kingdom 11 Continuous 350+450 CI Jersey, Cyprus 12 Continuous 400 or 425 Portugal 1 (ITU) 13 Continuous 400 Portugal 2 (Gagliardi) 14 Continuous 400 or 425 or 450 Ireland 1 (ITU) 15 Continuous 400 or 450 Ireland 2 (Gagliardi) 16 Continuous 400 or 450 Romania 17 Continuous 420 or 450 Austria 18 Continuous 425 Albania, Denmark, Estonia, Faroe Islands, Finland, Hungary, Iceland, Lithuania, Poland, Portugal 3 (300 085), Russia, Spain, Sweden, Switzerland, Norway 19 Continuous 425 or 450 Belgium, Germany, Luxembourg, Netherlands 20 Continuous 440 France 21 Continuous 450 Turkey

Telektronikk 3/4.1997 73 2 Ringing tones Functional definition The auditory indication to be presented to a user to indicate that applied to the called terminal or service. (This definition is con- a connection has been made and that an alerting signal is being sistent with ITU-T Recommendation E.182.)

Table of European Ringing Tones

Cadence Period Frequency Country

1 0,4 – 0,2 – 0,4 – 2,0 3 400 or 450 or 425 x 25 Ireland 1 2 0,4 – 0,2 – 0,4 – 2,0 3 400 x 25 Cyprus 2 3 0,4 – 0,2 – 0,4 – 2,0 3 400+450 Gibraltar, CI Jersey 4 0,4 – 0,2 – 0,4 – 2,0 3 400+450 or United Kingdom, Malta 450 x 25 or 425 x 1762/3 5 0,4 – 0,2 – 0,4 – 2,0 3 425 Ireland 2 (Gagliardi) 6 0,8 – 3,2 4 425 Lithuania, Russia 1 7 1,0 – 3,0 4 400 or 450 Norway 2 8 1,0 – 3,0 4 425 Cyprus 1 9 1,0 – 3,0 4 425 or 450 Belgium 10 1,5 – 3,0 4,5 425 Spain 11 1,83 – 3,0 4,83 425 Bulgaria 3 12 1,2 – 3,7 4,9 425 Hungary 13 1,0 – 4,0 5 425 Albania, Bulgaria 2, Croatia, Czech Republic, Denmark, Estonia, Faroe Islands, Finland, Italy, Lux- embourg 1, Norway 1, Poland, Russia 2, Slovakia, Slovenia, Yugoslavia 2 14 1,0 – 4,0 5 425 or 450 Germany 5 (Gagliardi), Greece, Netherlands 15 1,0 – 4,0 5 425 or 500 Switzerland 16 1,0 – 4,0 5 450 Luxembourg 2 (Gagliardi) 17 1,5 – 3,5 5 440 France 18 1,2 – 4,7 5,9 425 Iceland 19 1,0 – 5,0 6 400 or 425 Portugal 20 1,0 – 5,0 6 420 or 450 Austria 21 1,0 – 5,0 6 425 Sweden 1 22 1,0 – 5,0 6 425 or 450 Germany 3 23 2,0 – 4,0 6 400 x 16 or 450 x 25 Romania 24 2,0 – 4,0 6 450 Turkey 25 0,25 – 4,0 – 1,0 – 4,0 9,25 425 or 450 Germany 1 26 0,5 – 4,0 – 1,0 – 4,0 9,5 425 or 450 Germany 2 27 1,0 – 9,0 10 400 Sweden 2 28 1,0 – 9,0 10 425 Bulgaria 1 29 1,0 – 9,0 10 450 Germany 4 30 1,0 – 9,0 10 450 x 25 Yugoslavia 1

74 Telektronikk 3/4.1997 Annex 2 Examples of tone audits

Audit of tone ‘DIALLING’ 5 European mode for dial tone – Characteristics Continuous cadence, 425 Hz Single frequency. 1 Dial tone – Functional definition The auditory indication to be presented to a user to indicate that 6 International standards for dial tone a network connection is available and ready to receive call information and inviting the user to start sending call or service related information. (This definition is consistent with ITU-T Table of International standards for Recommendation E.182.) network generated dial tones

2 Position in HF model for dial tone Network generated tones This functional definition places the dial tone in the HF model Source Cadence Fre- Preference within the family of Feedback + System State OK and at the quency for any “Good” end of our characteristics continuum. It may also be update interpreted to suggest that the tone should be within the family of equip- Prompt Can. ment

ITU-T Continuous Single f, range Continuous 3 HF model characteristics for dial tone Rec. 400 – 450 Com- cadence at • Low frequency < 500 Hz E.180 bined, up to 3f single • Slow rhythm < 1/s (1f range frequency • Long period > 3 s 340 – 425, of 2f range 425 ± 15 Hz 4 Current European practice for dial tone 400 – 450) CEPT Continuous 400 – 450 Table of dial tones used in Europe T/SF 23 CEPT Continuous 425 ± 15 Cadence Frequency No. of Countries T/CS 20-15

Continuous 350 + 400 1 Continuous 350 + 440 1 Table of International standards for terminal generated dial tones Continuous 350 + 450 2

Continuous 400 or 425 1 Terminal generated tones Continuous 400 or 425 or 450 1 Source Cadence Frequency Continuous 400 or 450 1 Continuous 420 or 450 1 prETS 300 245-7 Continuous 425 or Continuous 425 14 (ISDN Terminals) or National National Tones Tones Continuous 425 or 450 4 Continuous 440 1 ETR 187 (General) Continuous 425 Continuous 450 1 GSM 02.40 (Dial tone N/A N/A 0,165 – 0,165 – 0,66 – 0,66 425 a is not normally required, but if it is provided) Continuous 425 ± 15 0,3 – 0,33 – 0,66 – 0,66 425 1 + a 0,2 – 0,2 – 0,6 – 1,0 425 1 ETR 294 (TETRA – N/A N/A no characteristics 0,7 – 0,8 – 0,2 – 0,3 425 2 defined) 0,2 – 0,3 – 0,7 – 0,8 425 1 0,2 – 0,3 – 0,7 – 0,8 425 or 450 b 0,2 – 0,8 – 0,7 – 0,3 425 or 450 b 0,25 – 0,3 – 0,7 – 0,8 425 c 0,25 – 0,75 – 0,75 – 1,0 425 c

Note: Letters are used for countries which report two or more different tone characteristics for the same function. See Part 2 for the fuller details.

Telektronikk 3/4.1997 75 7 Human factors issues for dial tone continuous to a Continuous tone, some to migrate from a dual to a single frequency and for others to change the frequency. Expectations A continuous low frequency tone of single or dual frequencies However, the HF evidence suggests that: Confusability • the differences within the range of single frequencies used is unlikely to cause any confusion; HF expertise would suggest that in the absence of the facility to compare two frequencies • the differences between the dual frequency and single frequency tones is unlikely to cause much confusion; • few users (may be as low as 10 %) would report a difference between two continuous single frequency tones, when the • the use of a discontinuous tone with a short period is the most variation of tones used is within ±25 Hz; likely source of confusion to Category A users. • more users (may be as many as 50 %) would report a difference between continuous multi-frequency tones and con- d) Do Something different tinuous single frequency tones; There is no human factors or technical reason for such a change • most users (may be as high as 95 %) would report a differ- to the public network service. Where there may be a require- ence between discontinuous single or multi-frequency tones ment to express a ‘corporate identity’ through a ‘unique’ dial and continuous single or multi-frequency tones. tone, any proposed tone or earcon would need to be tested for usability. Usability In practice most users of unfamiliar networks may assume that e) Do nothing any auditory signal presented on requesting connection to a net- The spread of characteristics of existing European dial tones is work can be taken to be the invitation to send call or service not large, and there are few confusion or usability issues. Where related information. these may be expected i.e. in the case of discontinuous tones and perhaps also in the case of dual frequency tones, the impact The alternative – Silence – would be taken to indicate no con- will be felt most by Category A users (travellers temporarily nection is available. using another network).

8 Review of options for dial tone 9 Implications of options for dial tone a) Use the HF Model Option a – Use the HF model If the dial tone is correctly located in the ‘Feedback/System No changes required. OK’ family, the characteristics should be located towards the lower frequency slower end of the range. If it should be located Option b – Use the Standards in the ‘Prompt can’ family, it should lie more towards the higher frequency fast end of the same characteristics. Change to a Continuous Cadence. EU In this context, if continuous is taken as the extreme of slow cadence, the HF model requirements are similar to the require- Greece (change cadence to continuous, and maybe change ments of the standards and also the existing European mode. NU tone for Paging) Italy (change cadence to continuous, and change Special Dial tone) b) Use the standards ETSI area The characteristics defined in the current ITU-T, CEPT and ETSI standards are all coherent with each other. They all define Bulgaria, Croatia, Slovakia, Slovenia & Yugoslavia (change a Continuous tone with differing tolerances on the acceptable cadence to continuous) frequencies. There are two EU and a number of other European Czech Rep. (maybe change cadence to continuous) countries that do not comply with the loosest standard ITU-T, these are identified in subclause 8.2.1.9. The key change re- Option c – Use the Mode quired for compliance with ITU-T is for countries with Dis- Change to a Continuous Cadence and a Single Frequency continuous Tones to migrate to a Continuous Tone. The (Based on Continuous at 425 Hz). changes required for compliance with the tightest standard are the same as for converting to the European mode. EU Greece (change cadence to continuous and maybe change f c) Use the European mode on some exchanges 450 to 425 and maybe change NU tone for Paging) The characteristics of the modal dial tone provide a good fit to Italy (change cadence to continuous; and change Special Dial the HF model, but there are a number of EU and other European tone) Countries with Dial Tones that do not precisely match these United Kingdom (change f from 350 + 440 to 425) characteristics; these are identified in subclause 8.2.1.9. The key Insignificant changes would be required in Austria, Belgium, changes required are for some countries to migrate from a Dis- Germany, Luxembourg, Netherlands, France, Ireland, Portugal

76 Telektronikk 3/4.1997 ETSI area 10 Recommendations for dial tone Bulgaria, Croatia, Slovakia, Slovenia, & Yugoslavia (change cadence to continuous) 1 To encourage the rapid harmonisation of the functional CI Jersey & Cyprus (change f from 350 + 450 to 425) use for dial tone and to encourage the cessation of the Czech Rep. (maybe change the cadence on some exchanges use of a similar tone to indicate other system functions, to continuous) e.g. positive indication. Gibraltar (change f from 350 + 400 to 425) Insignificant changes would be required in Rumania & Turkey

Option d – Do something different 2 To encourage the rapid change to the consistent use of a The implications of this course of action are dependent on just Continuous dial tone of single or dual frequencies. how different the ‘different’ is. At least it would imply extensive testing of the proposed tone or earcon.

Option e – Do nothing There are no additional costs in this course of action but bene- 3 To encourage the gradual change to a single harmonised fits will be foregone, particularly to terminal equipment manu- dial tone based on the Continuous tone of 425 Hz, when facturers and users. planned changes to the public networks make it possible.

4 To encourage National Regulatory Authorities (NRAs) to require new public network operators providing services which require the provision of a dial tone to comply with either their existing national dial tone characteristics or to provide a dial tone with a Continuous tone of 425 Hz.

5 To permit new dial tones based on the HF model that may assist the provision of a ‘Corporate Identity’ to Public Network Operators or Private Corporate Networks. NRAs should ensure that any new dial tones are properly regulated, are usable, are compliant with the HF model and are significantly distinguishable from network tones having other meanings.

Telektronikk 3/4.1997 77 The ergonomics of teleworking JOHN W. BAKKE

An inspection of the concrete opera- of more humane and democratic future In this presentation, selected themes in tion of any cooperative system shows societies [3, 4, 5]. the ergonomics of teleworking will be [...] that the physical environment is explored. Teleworking is potentially an an inseparable part of it. [1] Since the inception of the idea, a large interesting arena for studies of body of literature on teleworking has ergonomics, motivated by the specific A characteristic feature of teleworking is been produced, where a few predominant organizational characteristics of this way the geographical dispersion of an organi- themes have dominated the discourse: In of organizing work, as well as the sheer zation. This dispersion has profound and addition to a plethora of descriptive number of teleworkers. Although the distinct implications for the practice of accounts in the popular media, one finds concept of ‘teleworking’ covers a variety teleworking, compared to other work descriptions of technological develop- of work environments and work prac- arrangements. One set of implications ments, potentials and solutions, a large tices, it will be argued that there are cer- relates to the organization of work, like number of teleworking guides and hand- tain, well defined, ergonomic challenges co-operation, co-ordination and manage- books, and several case-studies, usually of teleworking. ment. Another set of implications relates pilot studies, in the field of organization to what might be called the ergonomics theory. In the first sections, a brief presentation of teleworking, the working environment of teleworking, and glimpses from challenges of dispersed work. A decisive background for the idea of ergonomics, organization theory, archi- teleworking is the development of infor- tecture and design will be given. This mation- and communication techno- provides a background for a discussion 1 Do you take your body logies, allowing, in principle, a large of the ergonomical challenges of tele- to work? number of work tasks to be performed at working, through an exposition of certain an arbitrary location. ‘Information work’ characteristics of this way of organizing The question “Why bring people to work, is considered especially appropriate for work. In this section, it will also be when they can work where they are?” telework, since it is easier to move infor- argued that studies of the ergonomics of may serve as a motto for the field of tele- mation (or ‘bits’) than bodies or products (tele)working may serve as a link to more working. Originally, the idea of tele- (or ‘atoms’ – see [6]). The interest in general themes in studies of organiza- working was motivated by a concern for teleworking is further motivated by tions, like gender, culture, and the pro- the oil-dependency of western intensified competition and a rapidly cesses of modernization. In a final sec- economies, and the potentials for reduc- changing competitive environment, new tion of the essay, implications for further ing road-based transportation [2]. Later, organizational forms and management research will be outlined. the theme of teleworking was incorpo- techniques, and the increasing pressure rated into the literature on office automa- on the family’s and the individual’s time tion, and into visions of the development schedules [7]. 2 Situating telework ‘Teleworking’ is not a well defined concept. While there are several examples, which doubtless do count as instances of teleworking, there is a large number of work situations that may – or may not – be counted as teleworking. For the purposes of this essay, an encompassing concept of teleworking will be deployed – mainly understood as work at a different location than the employer’s. This definition is considered beneficial, by allowing a wide variety of practices within a field of rapid development to be included in the analysis. A more precise definition may be obtained by specifying threshold levels for regularity and volume, technology deployed, and the existence of an agreement between employer and employee, while bearing in mind that such threshold levels are arbitrarily given [8, 9].

2.1 A brief history of telework

For a couple of decades, teleworking has been a theme in visions of the future of work. In a now classic study of the “telecommunication-transportation trade- off” [2], telecommuting was proposed as Figure 1 A number of work tasks may be performed equally well at a centralized workplace ... one solution for the oil-dependence of the

78 Telektronikk 3/4.1997 western economies. Aided by communi- While being internationally inspired, one employer and the employees in question. cations technologies, the argument went, also finds national characteristic traits. When discussing motivations, one may one might reduce road-based commuting For instance, local teleworking centres, distinguish between work-related argu- by working one or more days per week at or ‘telecottages’ have been proclaimed as ments, and non-work-related arguments home, or in a neighbourhood telework a Scandinavian solution, retaining the for teleworking. Further, one may centre, thereby reducing travel and traffic positive aspects of teleworking, like short assume that employers are more inte- congestion, in particular in urban areas. commuting travels and presence in the rested in the work-related implications of Later, the term teleworking was intro- local community, while avoiding perils teleworking, like increased productivity, duced as a general term for “ANY form like isolation or an unfavourable bargain- whereas employees are interested in both of substitution of information technolo- ing position relative to the employer [15]. work-related aspects like peace and quiet gies [...] for work-related travel” [7]. The for concentration away from a hectic concept telecommuting has, however, re- To a large extent, the literature on tele- office, and non-work-related aspects, mained in use as the general term, at least working has a pro-teleworking bias, related to commuting, or the manage- in the American context, while in Europe, often in mixtures of descriptions and pre- ment of the boundary between work and the concept teleworking has become scriptions. The majority of studies are non-work. accepted as the general term [8]. simple assertions of the possibility of teleworking, and interviews with happy Certain aspects of teleworking may be As indicated above, the theme of tele- and satisfied teleworkers and tele- conceptualized a priori, like the neces- working was soon incorporated in more workers’ managers and families. In addi- sity of overcoming the distance between elaborate visions of the future of work, tion, one finds a large number of case- the teleworker at the remote workplace, and the future of society at large. The studies from single or a few companies. and his or her colleagues, managers and motivations for establishing teleworking There are few comprehensive and even other relevant parties [7], or the chal- schemes went beyond the ‘substitution fewer comparative studies of this way of lenge of managing the boundary between for transportation’-argument, and tele- organizing work (for a discussion of the home and work, when working at home working was presented as a way of inte- literature on teleworking, and its often [16]. With teleworking as such a wide- grating work and family life [4], a way of dubious quality, see [8]). ranging phenomenon, there are, however, relocating work to rural areas [10], a way few ‘inherent’ characteristics of this way of offering peace and quiet for work The motivations for establishing tele- of organizing work. For instance, some groups [11], a way of capitalizing on working schemes vary considerably. It managerial strategies like ‘management investments in information technologies seems reasonable to assume that for tele- by walking around’ seems to be in- [12], and a way of slimming expensive working schemes to become successful, appropriate in teleworking situations. office space [13]. they have to some extent to benefit the This will, however, only be the case if parties involved – most directly, the the teleworker spends the majority of his In the study of the telecommunications- transportation trade-off [2], teleworking centres, or ‘business centres’, was given a prominent place, whereas home-based teleworking got less attention – although the authors referred to the media awareness of home-based work. It was the futurist Alvin Toffler who most force- fully formulated the ideal of home-based teleworking, in what he called the electronic cottage. The result, he proposed, would be a ‘home-centred society’. In later studies, home-based teleworking has come to the forefront, alone, or in combinations with office-based work and mobile teleworking (see [7] and [14], among other sources).

From its early beginnings, the discourse on teleworking has had an international orientation. The initial studies were made in the United States, but very soon the idea spread to other countries as well, while deploying the US, and later the British and Swedish experiences as in- spirations [8]. Now, information policies like the Bangemann Report promote teleworking, and the goal has been formulated to reach two million teleworkers in the European Union by the turn of the century [12]. Figure 2 ... or at a home-based office (Source: Information brochure from Telenor)

Telektronikk 3/4.1997 79 Sales - 4,3% Other - 56,0% occupations, the more important being education and research, managerial work, Office work - 4,6% office work, sales and accountancy (cf. Figure 3). Accounts - 3,9% The majority of the teleworkers had Programming - 2,2% themselves taken the initiative to establish their teleworking scheme (cf. Figure Management - 8,0% 4, where a prospective teleworker ad- vertises his services).

In order to bridge the distance between Journalism - 1,9% employer and employee, telephony is the most common method, followed, in order, by showing up in person, post, and telefax. Regarding technologies and ser- Education research - 19,1% vices deployed in their work as teleworkers, telephony proves to be the undisputed number one, followed by personal computers, the postal system, cellular telephony, answering machines and photo copiers [19].

The study demonstrates differences in communication behaviour between tele- Figure 3 Primary occupation of home-based teleworkers working employees and self-employed teleworkers. In general, the self- employed report a higher level of usage of most types of information- and communications technologies, compared to the employees. A notable exception is telephone conferencing, where the employees have a higher score. or her working time at a remote office, creation, and knowledge transfer to rural not if teleworking takes place intermit- areas. While stating that home-based In studies of communication patterns, the tently, or after ordinary working hours. teleworking already was the predominant lack of usage of information- and com- form, the report argued that this form of munications technologies is also of great 2.2 Teleworking in Norway teleworking would not provide a satis- interest. The survey of teleworking in factory work environment, according to Norway confirm that a number of tele- existing laws and regulations. Instead, workers do not deploy any communica- In Norway, teleworking was put on the telework centres are presented as a tion technologies at all. One implication political agenda in the early 1980s, in an preferable solution, providing a better might be to reconsider the conceptions of Official Norwegian Report on the future psycho-social work environment. A num- teleworking. Whereas teleworking is of telematics [15]. In this report, tele- ber of teleworking centres were estab- often depicted as a rational usage of working was primarily discussed as an lished, mainly on initiative of the authori- recent developments in information- and instrument for decentralization and job- ties, but they did not prove successful in communication technologies, these the long run [10]. findings indicate that teleworking might also be understood as a way of obtaining In spite of few formal teleworking peace, quiet and concentration, away schemes or company policies, home- from the workplace. based teleworking grew to some volume. In the mid-nineties some nine percent of In the mid-nineties a teleworking de- the working population perform parts of velopment programme was launched in their work-related activities at home, and Norway. This programme was initiated can therefore be called teleworkers in a and financed by the Research Council of broad sense of the word [17, 18]. The Norway, with additional support from majority of these teleworkers combine Telenor Research and Development. The work at home and work at another loca- core activity of the programme is a series tion. There is a fairly equal distribution of pilot experiments in a number of com- among men and women – 60 percent panies and institutions. Further, a number men and 40 percent women – and ap- of working groups are established, to dis- proximately two thirds of the teleworkers cuss and formulate policies on selected Figure 4 A prospective teleworker seeks employment are employed, one third self-employed. themes. The objective of the programme (Source: Aftenposten) The teleworkers represent a multitude of is to contribute to the further develop-

80 Telektronikk 3/4.1997 ment of teleworking, through syste- conceptualized as production techno- what management is about, and buildings mizing teleworking experiences, author- logies – were in the early phases of orga- influence behavior” [21]. ing a teleworking guide, and establishing nization theory (see also [24]). Later, the a teleworking forum [20]. scope of interest has broadened to In the development of Ford’s assembly include office work and service work, line factories, the building designs of and more detailed themes like strategy architect Albert Kahn were integral ele- 3 Science, good practice, formulation, the importance of the in- ments. The buildings were tailored to the and regulations formal organization ‘within’ the formal production processes, in particular to the one, and the role of management. flows of materials and cars under produc- The role the physical environment tion [30]. These assembly line factories plays in influencing organized activity One of the classic authors of organization were, however, best suited for long pro- [...] has received only modest atten- theory, Chester I. Barnard, formulated duction series of uniform products, since tion, particularly in the corporate what may be read both as a manifesto changing the production flow was a time- world. [21] and as a research program for the role of consuming process. the physical environment in organiza- A recurrent theme in studies of organizations. In The Functions of the Executive, Office buildings do not present the same tions, is the importance of the physical he distinguishes between on the one hand building requirements as factories do, working environment. This theme is the terms enterprise, business and opera- since the handling, transportation and treated in subject areas as diverse as tion, and on the other hand the term storage of information is comparably organization theory, architecture, design, organization, which is “reserved for that simple and flexible – even in paper-based and ergonomics. In this section, selected part of the cooperative system from offices. Office work may therefore be perspectives on the physical working which physical environment has been performed in a variety of settings. Cer- environment will be outlined, as a back- abstracted”. This process of abstraction tain types of office work have a long ground for a more detailed discussion of then forms a basis for more detailed in- tradition in western societies, like bank- the ergonomic challenges of teleworking. vestigations of relationships between “all ing, trade and commerce, and the work of Since ergonomics as a discipline does not aspects of the physical environment” and lawyers and notaries: “The notary was cover all substantial areas equally well, the organization, “as may be required for the inventor of the private office as we in this presentation, fragments from orga- the purpose in hand” [1]. (See also the know it” [31]. Until the eighteenth cen- nization theory, architecture and design quotation above.) tury, offices were usually integrated in will provide additional insights. the private home of the notary, lawyer or Several researchers follow Barnard’s trader, later, offices became located in 3.1 The materiality of the work- methodological dictum, whereas other designated office buildings, thereby replace research traditions present more abstract inforcing the physical demarcation be- and disembodied approaches, where tween the public and the private spheres. In organization theory, it is customary to organizations may be portrayed as regard F.W. Taylor [22] as one of the ‘nexuses of treaties’ [25], or through Early office buildings were similar to founding fathers. He performed several metaphors like ‘brains’, ‘cultures’, or apartment buildings, but gradually office studies of work processes, where he sys- ‘political systems’ (see the discussion in space became constructed on the factory tematically manipulated parameters like [26]). Theories of the ‘information so- model, with work in open spaces, and performance of work tasks and choice of ciety’, the ‘post-industrial society’, and with a supervisor centrally located, over- tools, and found substantial productivity the ‘post-modern society’ [27] may looking the employees. Later, the office impacts through what he called ‘scien- further encourage an understanding of building took the form of cellular offices tific management’. The productivity organizations – or entire societies – as arranged on each side of long corridors, impact was partially due to efficiency ‘immaterial’ [28]. This development with position and status manifested in the improvements, partially due to a reduc- towards a disembodied and immaterial- size of the offices, as well as the dimen- tion of shirking. He recommended ized theory is countered by other app- sions of the desk, type of chair and so on. changes in the division of labour, where roaches, emphasizing the materiality and Subsequently, the office buildings were managers should plan the production corporeality of work, organizations and supplemented with the landscaped open- process, whereupon the workers in detail social encounters. The bodily responses plan office (or Bürolandschaft) where the should follow the plans and instructions of work-related stress and repetitive employees typically were located accord- [23]. An even more thorough reorganiza- strain injuries demonstrate clearly that ing to workflow processes, the compart- tion of the work processes was instituted the lightness of information work may be mentalized office, and the flexible office through the development of the assembly quite unbearable [29]. [31, 32]. line, usually associated with the name of Henry Ford. Here, work tasks were sub- Workplace architecture and design repre- Around the turn of the century, office divided into small, repetitive elements, sent perhaps the most obvious forms of work changed radically: Office work and pacing and control were to a large physical working environment, and became a particular activity in a broaden- degree ‘delegated’ to the speed of the treaties on factory design and office ing range of businesses; a feminization of line (this production technology is paro- design provide valuable insights regard- the white collar work force started; and died in Chaplin’s film Modern Times). ing the way task performance, co-opera- the status of office work deteriorated.1 tion, and management are sought em- These examples, and numerous more, bodied or materialized in buildings: 1 It goes beyond the scope of this essay illustrate how central reflections on the “Influencing behavior is almost all of to even try to outline the mechanisms physical working environment – often interrelating these development trends.

Telektronikk 3/4.1997 81 The principles of scientific management an integrated theoretical perspective, and “meaningful occupation for the indi- were also applied to office work, with each of the elements of the conglomerate vidual employee”, demonstrating a differentiations within the category of provides valuable insights for workplace strong normative orientation where the office work as one result [31, 32, 33]. design – and thereby also for the design workplace is acknowledged as an arena and evaluation of teleworking as a work- with mutual social obligations, not From the late nineteenth century on, a ing environment. simply a meeting place, where work number of technological innovations efforts are exchanged for money. Finally, were introduced in the office, like the 3.2 Regulations of the working it is the objective of the Act to provide “a telephone, the typewriter, carbon paper, environment basis whereby the enterprises themselves and vertical filing systems [34], and can solve their working environment specialized office furniture was de- problems in co-operation with the organ- Ergonomics as a science, as well as orga- veloped to accommodate an increasingly izations of employers and employees” nization theory, architecture and design, diverse range of work tasks: Office desks (see also [42]). Thereby, a processual are all international in their orientation. with ‘scientifically’ designed filing space approach is chosen, where the legislation On the other hand, sets of rules and regu- were developed, as well as the sunken constitutes a framework for local activi- lations for the physical working environ- well desk with surfaces at different ties for the improvement of the working ment are typically anchored in local heights for the typewriter and for writing. environment. Regulations are formulated traditions and national legislation, like With office work as a full-time occupa- under the Act, and the Directorate of the respective worker protection acts and tion, not as one work task among others, Labour Inspection has a responsibility for corresponding regulations. In some the design of chairs became an important its enforcement. In spite of its enter- countries, the area of jurisdiction for topic, and adjustable chairs, designed for prising objectives, the Norwegian Work- worker protection regulations is rather a multitude of sitting postures – but with ing Environment Act does not apply to restricted, whereas other countries have different designs for different work “work performed in the employee’s widely encompassing regulations. Other statuses – became standard inventory in home”, although regulations may be for- national differences may be found in the offices [32]. mulated [43]. degree of specification of the respective regulations, and the means instituted to Whereas organization theory and work- ensure the regulations will be followed place architecture and design present 4 Ergonomical challenges (for a study of European approaches to comprehensive, overall perspectives on labour legislation, see [38]). of teleworking the relationships between the physical “The King shall decide whether and to work environments, work tasks, and There are certain attempts of creating what extent this Act shall be applicable organizations, their attention to detail is international regulations of teleworking, to work performed in the employee’s somewhat scarce and unsystematic. In like the ILO Convention and Recommen- home” contrast, ergonomics (or human factors, dation concerning home work, adopted The Working Environment Act as it is also called) is a workplace science by the International Labour Conference presenting detailed knowledge of work at its eighty-third session in 1996 [39, For teleworking to be possible, certain situations and topics like lighting, noise 40]. It may, however, still take some time work tasks must be capable of relocation and air quality, as well as working pos- before they will show concrete results, to other whereabouts than the traditional tures, task design, stress and fatigue [35]. due to the extensive process of ratifica- workplace.3 The crucial importance of Ergonomics as a science does also pre- tion and implementation in the respective information and communication tech- sent a normative orientation towards an countries. Within the framework of the nologies for teleworking, is that this fam- improvement of the interaction with the European Union, telework has been dis- ily of technologies allows a vast number working environment, and is therefore cussed as one instrument for the further of work tasks to be performed at an arbi- practically oriented: “Ergonomics has lit- development of a unified European trary location, and thereby being potential tle value unless it is applied” [35]. labour market [41]. candidates for teleworking. If teleworking solely related to the single work tasks in There are many specialities in the field of In Norway, the Worker Protection Act question, one could argue that it pre- ergonomics, like physical and cognitive was in 1977 replaced by a Working En- sented few generic working environment aspects of user-machine interfaces, work- vironment Act characterized by an challenges, if any: The respective work place design and workspace layout, phys- encompassing area of application. The tasks would present the same challenges ical and psychological working environ- objective of the Act is to “secure a work- as the not-relocated counterpart, and ment, and job design, selection and training environment which affords the should meet the same requirements, irre- ing (after [36]). In areas of ergonomics, employees full safety against harmful spective of the location where the work one may observe a shift in focus from physical and mental influences”2, tasks were performed. This line of rea- studies of thoroughly designed work- according to standards that correspond to soning would imply that the ergonomics places like a control room or a cockpit, “the level of technological and social of teleworking was no different from the towards studies of less structured work development of the society at large at any ergonomics of the similar work task in situations, like office work in general, not time”. Further, it is the objective of the another setting or location. While bring- solely typing or filing, or the functioning Act to secure “sound contract conditions” ing attention to the general issues of of work groups aided by computers (see [29] and the contributions in [37]).

Although ergonomics appears as a con- 2 The passages from the Working En- 3 An occupation or a job encompasses glomerate of particular studies, without vironment Act are quoted from [42]. numerous work tasks.

82 Telektronikk 3/4.1997 ergonomics to the field of teleworking there are relatively few traditions and means that when working at home, the studies, there is a major flaw in this argu- regulations for this area of work. Central teleworker somehow has to manage a ment: Teleworking is not solely about a questions are how teleworking organiza- boundary between home and work, by potential re-location of work tasks, but tions are designed, who are in charge of transforming or re-establishing what has also about a realization of this potential, the design, and the responsibilities in- been a territorial and time-based differen- where work is performed in settings with volved if the result proves to be an un- tiation, into a functional differentiation distinctive qualities, providing distinct satisfactory working environment. Top- between two classes of activities. One working environment challenges. ics like workplace inspection rights, strategy is to have one room as a desig- workplace insurance, regulation of work- nated work space, maintaining a geo- Teleworking represents a dispersion of ing hours, monitoring of efforts and effi- graphical distinction between work and the work organization, which, in turn, has ciency, and communication of expecta- non-work, while the practice of a strict implications for the functioning of the tions and results are but a few arenas time-schedule is another strategy for organization, like management tech- where responsibilities have to be re- upholding a home/work-boundary [16]. niques, relations to customers and col- assigned, and new procedures have to be leagues, and relations with family and established. The strategy of practicing standard work- friends. One may argue that the geo- ing hours means that belonging to the graphical dispersion of the organization 4.1 Working at home sphere of home or work may be deter- has to rely on a degree of trust between mined by the time of the day. This strat- the parties, and will also impose a need egy may be instrumental for managing The basic requirement for teleworking to for making work results more visible, as role expectations versus colleagues, cus- be possible, is that selected work tasks in the management technique of ‘man- tomers, family and friends – for instance may be performed at locations other than agement by objectives’ [7, 14]. An intro- regarding the timeframes during the day the employer’s. The home may, accord- duction of surveillance methods [44] it is appropriate to make or required to ingly, serve as a workplace, given that it may, however, turn out to be counter- receive a phone call. With very flexible fulfils fundamental demands of a work- productive, since surveillance may erode working hours, or with work interlaced ing environment – including adequate the crucial element of trust. National leg- with periods of non-work, the boundary work space, ventilation, lighting, and islation and collective agreements may management becomes even more com- correct working postures, as well as safe also regulate the usage of surveillance plex (in particular for the surroundings). storage, and access to the necessary methods [43]. information, tools and work materials Work proves to be a gendered practice, as [45, 46]. A common advice in tele- The dispersion of the work organization is seen in the distribution of men and working guides is to allocate one room to has also implications for the way posi- women in various occupations, their an office to ensure professionality, at tions are acquired and manifested in a respective positions in bureaucracies and least when teleworking is not an acci- decentralized and somewhat elusive corporate hierarchies, and in the distribu- dental or sporadic activity. Another organization [7, 8]. Teleworking may tion of wealth and obligations in their advice is to install a telephone line dedi- also have implications for the distribution work careers [47, 48]. Detailed studies of cated to work, both to ensure a profes- of work tasks in an organization, since businesses prove that gender differences sional attitude towards the surroundings, this way of organizing work often at work change over time, and that and to facilitate the income tax return. If implies that the teleworker performs numerous actors and activities both the teleworker expects customer visits to certain work tasks previously taken care uphold and change gendered perceptions the office, higher office standards may be of by colleagues, supervisors, and secre- and practices [49]. The home is another required, and the teleworker may even tarial services. thoroughly gendered arena, and one may find that zoning regulations will not presume that teleworking, and in parti- allow customer visits in residential areas. Teleworking has been the topic for a cular home-based teleworking, will be large number of predictions, recommend- understood differently by women and Home-based teleworking acquire distinct ations and warnings, but has so far turned men. One conjecture may be that for male characteristics through its location in the out to be less drastically different from teleworkers, work will permeate the home home: Home and work are two spheres ‘ordinary work life’ than imagined: Tele- sphere, while for female teleworkers, it with their own dynamics, but the two working has proved to be more home- will be the other way round.4 A Norwe- spheres are also closely interrelated in a based than expected, whereas telework- gian study of motivations for becoming number of ways. Our contemporary so- ing centres have not been able to fulfil teleworkers support this hypothesis, as a ciety is characterized by a fairly distinct the promises made; mobile or nomadic main motivation for male teleworkers separation, in space and time, between teleworking has grown in importance, was that they did not manage to complete home and work, and designated work and has got recognition as one of the their work tasks during ordinary working spaces and working hours function as main forms of teleworking; and finally, hours, while for the female teleworkers, a major mechanisms for the co-ordination teleworking has not turned out to be a main motivation was a wish to take better and synchronization of societies. This is full-time employment for the majority of care of family and children [18]. contrary to earlier societies, where work teleworkers, but an option for a share of was performed at home, without sharply the working week, or an additional or The teleworker does not only have to defined working hours. Teleworking, and supplementary activity. uphold and manage a distinction between in particular home-based teleworking, challenges the current distinction be- Teleworking is not a ‘pre-packaged’ way tween home and work, as the home is of organizing work – there are several 4 I owe this formulation to Øystein Gul- deployed also as a work space. This options and choices to be made, while lvåg Holter.

Telektronikk 3/4.1997 83 Working at more than one location may solve certain problems sketched above, regarding not fully equipped offices. On the other hand, this solution requires the teleworker both to maintain an acceptable working environment at each location, and to manage the transitions from work site to work site. The first mentioned concern may be solved by, as far as possible, reproducing the working environment at each (tele)working site. This choice may provide both functional and familiar working environments at each and every site, but may also prove to be both expensive and space consuming. Further, this choice may neglect the wish for a functional differentiation Figure 5 Teleworking arrangements may encounter legitimation crises between the various work sites: The (Source: [50]) mobile teleworker, for instance, may want to have a spartan workplace to allow for mobility. The mobile workplace may then be supplemented by a better equipped, stationary base. Al- though the strategy of replicating an entire working environment from site to site may not be desirable, access to, or the replication of selected elements, like the spheres of work and non-work, but a computer service department – all databases, e-mail accounts or computer must also manage the transition from one factors which may influence work task programs, may ensure functionality and sphere to the other. For ordinary work, quality and workplace performance. flexibility in the movement from one site the journey between home and work may to another. Working at multiple work serve as a very concrete rite de passage, If the teleworking activities are severely sites seems to require an increased whereas for telework, more symbolic affected by a spartan working environ- amount of planning and co-ordination to rites may be appropriate. In the literature, ment, one solution is to purchase the ensure productivity, accessibility and one finds advises like following the ordi- respective services from an independent availability. Working at multiple work nary workplace rhythms, or dressing as if provider, like a printing or copying ser- sites may also preclude the practice of going to a workplace outside the home. vice, or a third party computer service. deploying a personalized workspace, like Other solutions may be relevant, if the a shelf or a desktop, as a memory aid and 4.2 Working in a not fully ascetically equipped teleworking office is a work instrument. equipped office not the full-time workplace. With additional access to a fully equipped work The practice of working at multiple work environment, some of the challenges sites may also serve as a motivation for In the last few years, computers have mentioned may be solved by advancing not equipping any of the worksites fully, become a fairly common household or postponing certain equipment- or ser- by letting the various workplaces in sum appliance, as the prices of computer vice-dependent activities to the time of provide a fully functional working en- equipment have fallen sharply, and the day or the week when the appropriate vironment. This is perhaps most evident numerous applications and services have technology or service is available. in the so-called flexible offices, where a motivated the purchase of a home com- functional differentiation between puter. In some respects, the teleworking workspaces is expected, and where main office may therefore be as well equipped 4.3 Working at multiple work offices function more as meeting places as an ordinary workplace for the core sites or ‘touch-down’-offices (typically with activities or tasks of the teleworker. It fewer workplaces than employees), in seems, however, to be a fair assumption A common assumption in several treat- combination with home-based work and that few teleworking offices provide the ments of teleworking, in particular in work at the premises of customers or whole range of welcome equipment of a earlier writings discussing teleworking as clients [13, 51]. well established company, in particular a potential or a threat, is that the tele- in the case of mobile or nomadic teleworkers would spend the entire working working. Required equipment may be week in a remote setting. Studies of 4.4 Working in a dispersed almost everything from a first-rate copy- actual teleworking behaviour prove, organization ing machine, a colour laser printer, a however, that the full-time teleworkers shredder, or an able stapler. Further, the represent only a small share of the entire Working in an organization does not nec- teleworking office may suffer from the teleworking population, as the majority essarily imply close co-operation with absence of secretarial services, postal of teleworkers combines teleworking colleagues. A number of work tasks are services, the services of a receptionist, or with work at another site [17]. performed in relative autonomy and iso-

84 Telektronikk 3/4.1997 lation from the colleagues, like contact door indicates availability, while the which we use to define and make sense with customers or concentration-inten- sound of a conversation, even through a of our work and employment” [55]. The sive work. In such cases, the geographi- closed door, may indicate that a person is cultural embeddedness of an organization cal dispersion of the organization intro- busy and not to be disturbed. In a dis- is explicitly discussed in the literature on duced by teleworking, does not have to tributed organization, other clues have to organizational culture, where culture is present major challenges for the perfor- be interpreted, and other procedures have acknowledged as the ‘glue’ upholding mance of the singular work tasks. In the to be established to ensure the smooth the organization as a functioning, pro- teleworking literature, there is a common functioning of the organization. ductive unity: “The real business goes on observation that work task independence in the cultural network” [56].5 is positively correlated with telework- 4.5 Meanwhile, back at the ability [7, 52]. This observation may, in office Corporate architecture is a major arena turn, be taken as a recommendation for a for workplace symbolism (see the contri- decoupling of work tasks, in order to butions in [57]). It is common for major In the teleworking literature, one finds allow for teleworking. This recommenda- corporations to erect centrally located elaborate discussions of promises and tion may, however, be contrary to the office buildings as signs of importance potential pitfalls for actual and prospec- successful realization of the organiza- and wealth. Designated buildings and tive teleworkers and managers, together tion’s goals, or detrimental to the func- canteens for ‘white collar’ and ‘blue col- with advises and guidelines for the tioning of the organization as a social lar’ workers serve as instruments for in- successful implementation and practice system, with consequences for organiza- group communication and identification, of telework. Teleworking studies do, tional learning, and organizational and for upholding status hierarchies. Fur- however, in less detail address how the belonging. One may therefore argue that ther, the size and location of offices are introduction of a teleworking arrange- this approach is rather short-sighted, by common indicators of status in the corpo- ment affects those in an organization considering telework or teleworkability a rate hierarchy, an anteroom with a secre- who are not part of the teleworker- goal in itself, not as means for improving tary demonstrates power, and the possi- manager dyads. Those remaining at a the quality of work and the quality of life. bility to refurbish and personalize an common office, either by choice or by office is an acknowledgement of one’s not being allowed to telework, will be Work tasks relying on extended co-oper- importance for the organization (see the affected by a teleworking arrangement in ation with colleagues will typically be discussion of ‘power offices’ in [31]). several ways: Their work tasks may more profoundly affected by a geo- change as a complement to the changes graphical dispersion, since co-operative With teleworking, the corresponding introduced at the teleworking sites, as practices, as well as conceptions of co- activities and processes will follow discussed above. Further, they may have operation, typically rely on an assump- somewhat different rules than at a co- to compensate for, or cover work tasks tion of co-location. Distributed co-opera- localized workplace, in particular when a which otherwise would have been per- tive work presents challenges for meet- substantial portion of the working week formed by the teleworker if she or he had ings and discussions, for the distribution is spent outside a common work site, or been physically present, like spontaneous and access to information, for the co- when a flexible office solution with ‘non- meetings or certain enquiries from cus- ordination of activities, for organizational territorial’ offices has been adopted. tomers or clients. With colleagues widely learning and the formation of shared While for instance career development scattered, the in-house workers may also opinions, and for joint access to the work may be manifested in the physical re- be affected in a similar way as the tele- material and work tasks in question. In location from a smaller to a larger office, workers, regarding the changed frame- the vivid research areas of computer-sup- or from one wing of a building to another work for knowledge development and ported co-operative work (CSCW) and – in a telework setting, other markers of -transfer, workplace socialization, and distributed work, these challenges are group affiliation and social status have to the development of company affiliation discussed, and technologies for solving be developed. and company loyalty. Finally, the social them are developed like document man- atmosphere at a workplace may be af- agement systems and screen-sharing and For corporate success, deliberate fected by teleworking, in particular when joint editing technologies (see the ‘impression management’ is recom- there is an extended telework pro- readings in [53] and [54], among several mended, including proper clothing, po- gramme, with only a small share of the other sources). liteness and attentiveness, and non-verbal employees present at a given time. impression management [58]. In a geo- Numerous aspects of co-operation and graphically dispersed organization, the socialization in an organization are af- 4.6 Workplace symbolism ‘presentation of self in everyday life’ will fected by teleworking, as physical loca- follow different rules, than “within the tion and other physical clues provide The workplace is a site both for pro- physical confines of a building or a valuable information for the functioning ductive work and for the production and plant” [59]. Impression management, and of an organization. The importance of circulation of meanings and interpreta- the corresponding interpretations of cor- this kind of information is acknowledged tions, inseparable from ‘productive when considering the multitude of ac- work’. All economic activities are in- tions being performed during an ordinary herently cultural phenomena, because work day: Proximity of workplaces is “the processes of organization and pro- 5 Here, ‘culture’ is typically understood usually an indication of co-operation or duction [...] are the subject of cultural as a variable, subject to manipulation belonging to the same work-group, light change and reconstruction [and] ideas, in the pursuit of organizational pro- in an office or a coat on a hanger may images values provide us with shared ductivity (for a discussion and critique, indicate that a person is present, an open frameworks, assumptions and moralities see [55]).

Telektronikk 3/4.1997 85 As indicated above, telework is one of potential teleworkers, one also example of what may be called de- encounters a diversity of motivations, standardized or detraditionalized work, competences, and need for support. where the established regime of workplace regulation is Teleworking represents challenges for becoming fragmented, the way workplace influence, in all its and standardized forms, is performed, due to the geo- working con- graphical dispersion of the work organi- ditions zation.7 Further, the geographical dis- are persion has implications for how opinions about workplace policies are porate formed, a factor of major importance behaviour, for the practice of workplace has to rely on dif- democracy, usually taken care of ferent means of in informal meetings around cof- expression and under- fee machines, water coolers, and standing when relating to col- luncheon tables. leagues directly, or mediated via communication technologies [60, 61]. Tele- working has been described as “a great being 5 Implications for future equalizer”, offering equal work options replaced by for everybody, regardless of age, gender research local or individ- or disability [62]. The advent of modernity increasingly ual agreements (cf. tears space away from place by foster- [66] and the discussion of When relying entirely on mediated coming relations between ‘absent’ others, modernization processes in munication, it is possible to stage the pre- locationally distant from any given sit- [67]). sentation of one’s self rather freely, as is uation of face-to-face interaction. epitomized in the New Yorker cartoon “On [67] the Internet, nobody knows you’re a dog” The absence (relatively speaking) of (quoted in [63]) or in the car bumper traditions and regulations requires a A main objective of this presentation has sticker text: Work naked: Telecommute! number of choices to be made by the been to elaborate certain working en- parties involved. With several degrees of vironment challenges of teleworking. As 4.7 Workplace design and freedom, few paradigmatic models, and a result, the presentation may be read as a workplace democracy few standards and regulations, the catalogue of problematics and stumbling prospective teleworkers or teleworkers’ blocks for the design of dispersed work employers may face what may be seen as arrangements, or even as an argument Portability of work tasks is a prerequisite an insurmountable complexity of against distributed work. A problem- for teleworking to be possible, but tele- choices, or they may settle for solutions orientation is, however, inherent in working is not solely about work task inappropriate for their organizational discussions of ergonomical challenges, performance at an arbitrary location, as needs. The many options and choices in whereas the pleasures and productivity telework also relates to the frameworks teleworking also actualize questions impacts of teleworking belong to another within which work tasks are performed, about how, and by whom, a correspond- framework of discussion. like the contractual arrangement between ing framework should be designed, to the teleworker and the employer, or the what extent this framework should be A more appropriate reading of this teleworker’s influence on his or her detailed, and for whom it should be presentation is that teleworking is to be working environment and working con- applicable [39, 43]. understood in contrast to, or as an ditions. All work is embedded in frame- amendment to, specific work practices works of labour legislation, agreements As recommended in the handbooks and and working life institutions, and that the (collective and/or individual), traditions guidelines, several teleworking pro- ergonomical challenges of telework have and norms. For ‘ordinary work’, this grammes start with a pilot project to try to be compared with the ergonomics of framework is fairly extensive, whereas out technological and organizational existing work life situations. Telework- for telework, the corresponding set of solutions and to develop a contractual ing as a relatively new way of organizing regulations is poorly developed. framework for teleworking. Typically, work, has to prove its relevance and the next steps will be an evaluation of the attractiveness, while existing work life Teleworking has by some authors been pilot project, and a decision regarding the arrangements have an authority through seen as an opportunity for a radical re- future of teleworking in the organization. their sheer existence and familiarity. organization of working life: Advancing from a pilot project towards “[T]elecommuting could be that bit of full-scale teleworking, one encounters reasoning that will allow managers to questions of the scale and scope of tele- 6 Here, one may add the question: Will once again take control without the use working in an organization: Is telework- having a home suitable for teleworking of a whip and bullhorn” (sic) [64]. This ing for the selected few, just like an addi- become a requirement for being type of quotations demonstrates clearly tional benefit, or shall the majority of the employable? the importance of addressing normative employees be eligible as candidates for issues regarding the future development 7 In this respect, teleworking is quite teleworking.6 When extending the basis of teleworking [65]. similar to shift-work.

86 Telektronikk 3/4.1997 There are numerous technological and working studies is the privileged oppor- An interest in the ergonomics of tele- organizational responses to the ergonom- tunity for studying a new organizational working draws attention to the materi- ical challenges of teleworking outlined in form in the making. ality of work life, like the roles of tech- this presentation, like the development nologies and workspace architecture for and introduction of appropriate informa- The establishment of a regulatory frame- co-operation and work task performance, tion and communication technologies work for teleworking will impose certain and serves as a theoretical impetus for [68], the establishment of communication constraints on the multitude of choices in acknowledging the importance of the practices and electronic meeting places teleworking. The development of domi- minutiae of work life practices. equally accessible for everybody in an nant practices through a combination of organization, or the introduction of evolutionary processes and the impact of The malleability of work arrangements is meeting days where everybody is ex- teleworking handbooks and teleworking by no means confined to the establish- pected to be physically present [69]. It development programmes will in a ment of teleworking schemes – all work has not, however, been the intention in similar way, although far less coercive, arrangements are results of a multitude of this presentation to outline potential serve to make the number of choices less choices and decisions, but for existing solutions in any detail. overwhelming. There are important re- work arrangements they are rarely as search tasks in the monitoring of these open-ended and explicitly discussed. In With teleworking as a fairly new way of emergent practices, in the evaluation of the on-going transformation of economic organizing work, there is an abundance teleworking programmes [20], and in a life, one finds that every aspect of work of topics for further research. It is of conceptual differentiation between is increasingly more exposed to delibera- great importance to systematize existing various forms of what today is called tion and competition, and work tasks research, to monitor teleworking prac- teleworking. may be re-configured, re-assembled, or tices and experiences – with special obliterated, and where teleworking is but attention to gendered practices and ex- 5.2 The ergonomics of tele- one of many workplace and market inno- periences – and to raise normative issues working and the future vations (see the contributions in [72, to contribute to the further development of work 73]). Teleworking is in this respect of teleworking as an attractive way of interesting, since it challenges concep- organizing work. In particular, there are Ergonomics as a discipline is born from a tions of the localization of work, and research challenges in following tele- concern for the body in worklife situa- proves that work and non-work are not working arrangements beyond pilot tions, where unsatisfactory working con- two entirely separable worlds. Further, experiments, since one may assume a ditions are met with the ‘revenge of the when comparing with other ways of ‘honeymoon effect’ in the initial phases. body’ through work-related illnesses, and organizing work, teleworking demon- One may also assume that experiences a reduction of productivity and a deterio- strates most clearly the importance of will differ, when teleworking becomes an ration of the quality of work life. Corpo- communication in organizations, with the option for others in an organization than reality is also central in studies of culture associated competences, cultures and the most enthusiastic teleworking and communication emphasize the body technologies. pioneers. Further, there are important as a locus for intention and meaning, and research topics regarding the develop- the richness of face-to-face communica- ment of even more ‘elusive’ organization: Wherever work is performed, one 6 References tions, with distributed work as a domi- does take one’s body to work [60, 61, 70]. 1 Barnard, C I. The functions of the nant work practice. Further studies have to be made about the executive. Cambridge, Harvard Uni- usage of mediated communication in versity Press, 1968. 5.1 Developing teleworking teleworking and other distributed work- regulations, practices and life situations, of media choices, and 2 Nilles, J et al. The telecommunica- concepts finally, about the development of com- tions-transportation trade-off. New munication technologies, and the appro- York, John Wiley, 1976. As pointed out above, the phenomenon priation or ‘domestication’ of these tech- of teleworking is somewhat ill-defined, nologies through the development of 3 Kraut, R E (ed.). Technology and the and it has been proposed to abolish the appropriate communication strategies [71]. transformation of white-collar work. term completely. The concept has, how- Englewood Cliffs, Larence Erlbaum, ever, its acknowledged benefits: “Despite A predominant theme in this presentation 1987. its lack of precision, the word telework has been the malleability of work has acquired a potent symbolic value” arrangements, which has been illustrated 4 Toffler, A. The third wave. New [8]. The problems with establishing a through glimpses from the history of York, Pan Books, 1981. clear-cut definition of teleworking may work, and, in more detail, through a be lamented as an indication of the presentation of the numerous choices to 5 Julsrud, T E. Teleworking – the immature status of the teleworking re- be made for a ‘detraditionalized’ way of vision. Telektronikk, 92(1), 16–25, search, but may equally well be taken as organizing work like teleworking. 1996. an indication of the multitude of choices Choices represent both potentials and in the establishment of teleworking prac- burdens, and the lack of regulations and 6 Negroponte, N. Being digital. Lon- tices – concordant with the characteriza- paradigmatic examples of teleworking don, Hodder & Stoughton, 1995. tion of teleworking as a destandardized means that the prospective teleworking or detraditionalized work practice. organization is forced to spend some Hence, the other side of the (alleged) effort specifying the work arrangement. immature status of teleworking and tele-

Telektronikk 3/4.1997 87 7 Nilles, J. Making telecommuting hap- 21 Seiler, J A. Architecture at work. 38 Nyström, B. EG och arbetsrätten. pen. New York, Van Nostrand Rein- Harvard Business Review, Sep-Oct, Stockholm, Publica, 1992. hold, 1994. 111–120, 1984. 39 ILO. Home work. Report V (1). Inter- 8 Huws, U et al. Telework : Towards 22 Taylor, F W. The principles of scien- national Labour Conference 82nd the elusive office. Chichester, John tific management. NY, Norton, 1967. Session 1995. Geneva, International Wiley, 1991. Labour Office, 1994. 23 Rose, M. Industrial behaviour. Har- 9 Bakke, J W. Å telle fjernarbeidere. mondsworth, Penguin, 1982. 40 Haakonsen, H. Taus arbeidsgiver- Kjeller, Telenor Research & protest. Arbeidsliv, No. 11, 4–5, 1996. Development, 1996. Unpublished 24 Rabinbach, A. The human motor. document. Berkeley, University of California 41 Actions for the stimulation of trans- Press, 1990. border telework and research co- 10 Hetland, P et al. Nært, men likevel operation in Europe. Brussels, Euro- fjernt. Stavanger, Rogalandsforskn- 25 Aoki, M et al. (eds.). The firm as a pean Commission DG XIII, 1996. ing, 1989. nexus of treaties. London, Sage, 1990. 42 Gustavsen, B, Hunnius, G. New pat- 11 Spinks, W. Satellite and resort terns of work reform. Oslo, Uni- offices in Japan. Transportation, 18, 26 Morgan, G. Images of organization. versitetsforlaget, 1981. (4), 1991, 343–363. London, Sage, 1986. 43 Jakhelln, H. Fjernarbeid. In: H Jak- 12 Club de Bruxelles. The future of the 27 Kumar, K. From post-industrial to helln and J W Bakke (eds.). Fjernar- information society (the ‘Bangemann post-modern society. Oxford, Black- beid og arbeidsrett. Kjeller, Telenor Report’). Brussels, 1994. well, 1995. Research & Development, 1996.

13 Becker, F, Steele, F. Workplace by 28 Diani, M (ed.). The immaterial 44 Booth, W. Big brother is counting design. San Francisco, Jossey-Bass, society. Englewood Cliffs, Prentice your keystrokes. In: S R Corman et 1995. Hall, 1992. al. (eds.). Foundations of organizational communication : A reader. 14 Kugelmass, J. Telecommuting : A 29 Grandjean, E. Ergonomics in com- NY, Longman, 75–76, 1990. manager’s guide to flexible work puterized offices. London, Taylor & arrangements. NY, Lexington Francis, 1987. 45 Rosenbaum, A. The complete home Books, 1995. office. Harmondsworth, Viking Stu- 30 Brunnström, L. Den rationella fab- dio Books, 1995. 15 NOU. Telematikk : Teleutvalgets riken. Umeå, Dokuma, 1990. utredning II. Oslo, Universitetsfor- 46 Kanarek, L. Organizing your home laget, 1983. (1983:32.) 31 Pélegrin-Genel, E. The office. Paris, office for success. Harmondsworth, Flammarion, 1996. Plume, 1993. 16 Nippert-Eng, C E. Home and work. Chicago, The University of Chicago 32 Forty, A. Objects of desire. London, 47 Mills, A J, Tancred, P. (eds.). Press, 1996. Thames and Hudson, 1986. Gendering organizational analysis. Newbury Park, Sage, 1992. 17 Bakke, J W. Fjernarbeid – en visjon i 33 Lupton, E. Mechanical brides. NY, forandring. In: J Moe (ed.). Fra Cooper-Hewitt National Museum of 48 Kvande, E, Rasmussen, B. Nye kvin- Kardemomme by til cyberspace. Design, 1993. neliv : Kvinner i menns organisa- Arbeidslivets utfordringer og sjoner. Oslo, AdNotam, 1990. muligheter i informasjonssamfunnet. 34 Yates, J. Control through communi- Trondheim, Norges Tekniske Viten- cation. Baltimore, The Johns Hop- 49 Kwolek-Folland, A. Engendering skapsakademi, 1995. kins University Press, 1993. business : Men and women in the corporate office 1870-1930. Balti- 18 Bakke, J W, Bergersen, E. Tele- 35 Wilson, J R, Corlett, E N (eds.). more, Johns Hopkins University working : The Norwegian experi- Evaluation of human work. London, Press, 1994. ence. Kjeller, Telenor Research & Taylor & Francis, 1995. Development, 1996. 50 Pursell, C. White heat. People and 36 Galer, I (ed.). Applied ergonomics technology. London, BBC Books, 19 Bakke, J W. Communication patterns handbook. London, Butterworths, 1994. in teleworking. In: K Nordby et al. 1987. (eds.). Human-computer interaction. 51 Raymond, S, Cunliffe, R. Tomor- London, Chapman & Hall, 1995. 37 Baecker, R M et al. (eds.). Readings row’s offices. London, E & FN Spon, in human-computer interaction. San 1997. 20 Bakke, J W et al. Teleworking de- Francisco, Morgan Kaufmann Publ., velopment programme in Norway. 1995. 52 Shaw, L. Telecommute! Go to work Kjeller, Telenor Research & De- without leaving home. NY, John velopment, 1997. Wiley, 1996.

88 Telektronikk 3/4.1997 53 Greif, I (ed.). Computer supported 67 Giddens, A. The consequences of cooperative work : A book of read- modernity. Cambridge, Polity Press, ings. San Mateo, Morgan Kaufman, 1990. 1988. 68 Eldib, O E, Minoli, D. Telecommut- 54 Engström, Y, Middleton, D (eds.). ing. Boston, Artech House, 1995. Cognition and communication at work. Cambridge, Cambridge Uni- 69 Sakshaug, U et al. Håndbok i fjern- versity Press, 1996. arbeid. Oslo, Norges forskningsråd, 1997. 55 Salaman, G. Culturing production. In: P Du Gay (ed.). Production of 70 Turner, B. The body & society. Lon- culture / cultures of production. Lon- don, Sage, 1996. don, Sage, 1997. 71 Haddon, L, Silverstone, R. Telework- 56 Deal, T, Kennedy, A. Corporate cul- ing in the 1990s : A view from the tures : The rites and rituals of cor- home. Brighton, University of Sus- porate life. Harmondsworth, Pen- sex, 1993. guin, 1988. 72 Olberg, D (ed.). Endringer i arbeid- 57 Gagliardi, P (ed.). Symbols and arti- slivets organisering. Oslo, FAFO, facts : Views of the corporate land- 1995. scape. Berlin, Walter de Gruyter, 1990. 73 Hesselbein, F et al. (eds.). The organization of the future. San Francisco, 58 Rosenfeld, P et al. Impression Jossey-Bass, 1997. management in organizations. Lon- don, Routledge, 1995.

59 Goffman, E. The presentation of self in everyday life. Harmondsworth, Penguin, 1990.

60 Barry, D, Hazen, M A. Do you take your body to work? In: D M Boje et al. (eds.). Postmodern management and organization theory. Thousand Oaks, Sage, 1996.

61 Stohl, C. Organizational communication. Thousand Oaks, Sage, 1995.

62 Ingham, C. New work options. Lon- don, Thorsons, 1996.

63 Turkle, S. Life on the screen. Lon- don, Phoenix, 1997.

64 Mahfood, P E. Homework . How to hire, manage and monitor employees who work at home. Chicago, Probus, 1992.

65 Bakke, J W. Telework, participation and workplace democracy. In: P John Willy Bakke is Research Scientist at Telenor Jackson and J van der Wielen (eds.). Research & Development, Kjeller. He is currently completing his doctorate at the Depart- Amsterdam ’97. Second international ment of Sociology and Human Geography, University workshop on telework. Tilburg, Work of Oslo. and Organization Research Centre, 1997. e-mail: [email protected] 66 Beck, U. Risk society. London, Sage, 1992.

Telektronikk 3/4.1997 89 Human factors of videotelephony KNUT NORDBY

Experience with videotelephony in Even as late as the 1880s, the French dialling and pushing buttons (at first only Norway and elsewhere has clearly futurist writer Albert Robida in his book for local calls, later also for long distance shown the need for operating pro- “Le Vingtième Siècle” envisaged a and international calls), but even these cedures and controls to be as simple mirror on the wall as the display for his were not beyond the capabilities of most and easy to use as possible. It was de- ‘Telephonoscope’. This metaphor was seven year olds. Today, however, there is cided to incorporate human factor also used by Jules Verne for the ‘cinema a host of new services which may be aspects to ensure a high degree of telephone’, a combined sound telephone accessed from a telephone, but they usability from the very beginning. and ‘tele-cinema’, in his 1889 short story involve struggling with long numbers, “In the Twenty-ninth Century: The Day arbitrary service codes and cryptic line A videotelephone is something much of an American Journalist in 2889”. signals. more than just a sound telephone with a video picture added. In the video- The possibility to speak to someone not The advanced technology of today’s tele- telephone we have to adopt new ways present had been realised in 1876 when phony, and other areas, has resulted in of communicating with other people Bell and Gray invented the telephone, increased user-problems and has raised and we must learn to operate several which soon became an accepted form of many interesting human factors issues. new controls, the results of which are communication. It was not until 1927, Although these issues have been dealt not always obvious to us. This poses however, that Herbert I. Ives at the Bell with in more detail elsewhere, they hold many new and different problems Laboratories made the first telephone call important implications for videotele- from the point of view of human fac- with a picture between Washington and phony. tors compared to the much simpler New York, although a picture was only sound telephony. This article is based transmitted one way. Technology had both on well established and accepted 1.2 Knowledge in the head and proceeded so far that it was now possible practices in the fields of cognitive and in the world to see the person with whom you were human factors and on experience talking as well. gained in our trials. To use any telecommunication terminal or system, the user must have a concep- Finally, in 1936 in Germany, full two- tual model of it and of the service it pro- way picture and audio communication vides. The model must be simple to com- 1 Visual distance was established between the main post prehend and it must cover the essential offices in Berlin and Leipzig [1], demon- communication features of the service. Some models are strating the feasibility of picture tele- derived from explicit knowledge in the To be able to see and talk to a person phony, or to use the accepted modern world, provided by the device itself, by who is not present is one of man’s oldest term, videotelephony. the instructions and by the physical con- dreams, the beginnings of which go from straints. Other mental models are derived way back to the mist of mythology and Lately, there have been a number of from knowledge in the head and are fairy tales. Reflections in fluids, crystal videotelephones and videotelephone pro- created in our minds from our interactive balls and magical mirrors (as the mirror jects around the world, with rather varied strategies and general rules for interac- on the wall in Snow White) are examples results, but no videotelephone has had tion, analogies, general knowledge, past of fanciful metaphors for ‘audio-visual’ any commercial success so far. It is very experience with similar devices, etc. distance communication which have easy to attribute this to such factors as become part of our folklore. cost, availability, usability and appro- All control procedures must be consistent priate applications. The topic here is to with this conceptual model and all the take a closer look at some important visible and audible parts of the terminal usability factors. must reflect the terminal’s current state in a way that is consistent with the con- 1.1 Basic human factors ceptual model.

Very complex machines – and the tele- The user needs simple, intuitively under- phone system is the world’s largest and stood and easily remembered mental most complex machine – must have models, maps and metaphors of what the simple basic operating procedures which terminal can do. Thus, the conceptual are consistent with human physical, per- model does not have to be an accurate ceptual and mental capabilities [2]. This representation of what actually goes on is essential if the end-user is to be able to inside the terminal. It is the end result of interact confidently and directly with the the service provided by the system that is equipment. of interest, not how it is performed by the system: e.g. it is irrelevant to the user In the beginning, all telephone calls were how the picture is transmitted, as long as handled by operators and the only thing the picture is displayed on the screen. the caller had to do was to lift the re- The mental model or the metaphor here ceiver off-hook and ask to be connected is simply that of a ‘distance mirror’. to the desired party. As telephone tech- Figure 1 Alleviating absence by use of the ‘Telephono- nology developed, it was necessary to Very often there is a clear discrepancy scope’. A mirror is envisaged as the display screen. adopt new call set-up procedures, such as between the conceptual model of the (From Albert Robida ‘Le Vingtième Siècle’, 1880s)

90 Telektronikk 3/4.1997 designer (if there ever was one) and the conceptual model formed by the user. The designer is in a totally different position from the user since the designer usually has a number of constraints govern- ing the design work: economy, deadlines and tight schedules, compatibility with existing systems, design traditions, corporate identity, marketing department requests, current fashions and trends, cultural traditions, a too detailed knowledge of how the thing actually works, etc.

The user, on the other hand, will have mental models that are based on: realistic or unrealistic expectations of what the device should do, previous experience with similar devices, no detailed knowledge of what goes on inside, sales talk and brochures, myths and rumours, or just wishful thinking. It is important to determine the user’s conceptual model of the product at the earliest possible stage in the design process and design to comply with it. The domestic video recorder (VCR) is a good example of a particularly poorly designed product user interface which has given problems to most users.

1.3 Simplify the task

Complex tasks requiring many control procedures are difficult to manage and remember by most people. Wherever Figure 2 The first design concept for a fully working videotelephone from ALCATEL developed in possible, tasks should be simplified by co-operation with Norwegian Telecom Research (now Telenor R&D) in 1987. Observe the camera lens changing the structure and minimising under the display screen. This position was chosen to improve eye-contact the need for problem solving and planning ahead. No demand should be made on short-term memory as this is a limit- ing factor with most people. Technology should be used to make things visible, relieve short-term memory and give feedback. sensed by human observers. To be able If, for example, you want to call the to use modern electronic equipment (e.g. wake-up service to catch a flight, but by Simple in this context means that proce- videotelephones, computers, etc.) it mistake you dial the code for diverting dures should be; intuitive (i.e. the natural should not be necessary to understand incoming calls to another telephone thing to do); consistent (the same action their internal workings. (there is no feedback to warn you) or, always gives the same result); easy to having dialled the correct service code remember by infrequent users; easy to Unfortunately, many telephone services, you misdial the time or the date, there is learn by first time users; self-explana- both those offered in the public network no feedback here either. In other words; tory; require little or no instruction; and and in private business exchanges, make you have no way of knowing if you will should give clearly visible results. These use of cryptic procedures that are diffi- be awakened at the right time – or even requirements are difficult to fulfil on cult to remember. To initiate a service the right day. many complex terminals, but they should you first have to dial the exchange, then a constitute a design goal to strive for. ‘*’ to initiate the command sequence, There is a sinister anecdote told about a then an arbitrary service code number man who wanted to make an external call 1.4 Make things visible (e.g. ‘63’), then another telephone num- from a strange office he visited. Assum- ber or a time and date, and finish off the ing that an ‘0’ would get him an external sequence by dialling ‘#’. Normally, no connection (in fact it was ‘9’), he dialled The user cannot sense directly what goes feedback is given during the process and the number. By pure coincidence, the on inside electronic equipment, unless as long as valid numbers are entered number he dialled reprogrammed the sys- the equipment displays information about there is nothing to warn you of a mistake. tem to broadcast to all the telephones in its internal state in a form that can be

Telektronikk 3/4.1997 91 1.6 Constraints

One very important way to achieve a simple and user-friendly design is to utilise natural and artificial constraints. The user should get the feeling that there is only one possible thing to do – i.e. the right thing.

A natural constraint is based on what the user believes to be the only natural and logical course of action, i.e. when there is only one light switch for the room lights, when there is only one way out, when there is only one button to press, etc.

Artificial constraints means that the device is designed so that one action cannot be executed before another, essential, action has been carried out, e.g. you cannot dial before you have a dialling tone and you cannot get the dialling tone before you lift the handset off-hook, etc.

1.7 Errors Figure 3 The latest Tandberg ISDN portable videotelephone with folding flat display screen, the Tandberg Vision 600 If errors can be made, people will invariably make them. All terminal equipment and systems should be designed to support rather than fight the user. Never blame the user. Nothing is achieved by harassing the user who may already be painfully aware that an error has been the system and also to turn them all to When making new metaphors, it is made. If errors are easy to make, the loudspeaking mode. There was nothing important that they mimic an essential designer is nearly always at fault. Design on his terminal to indicate to him that feature of the process in such a way that to avoid errors, but expect that errors will every sound he made was heard in every it supports the user’s naïve concept of the always be made and design to make it office in the building. process. easy to correct them. It is essential that metaphors and mental 1.5 Cognitive maps, models and Errors are costly in all systems. They maps make the user understand the rela- metaphors waste time, money and resources, they tionships between intentions and possible can endanger people’s health and life, actions; between any actions and their In order to understand how to use equip- and they are costly to correct. To avoid effects on the system; between the actual ment and procedures, we often devise our errors through inherent sound design system state and what the user perceives own internal “maps” and “models”, often should be the prime concern in all by the visual, auditory and haptic senses; naïve and simplified, according to our design. and between the perceived system states perception of how the equipment and and the user’s intentions, needs and procedures work. Such mental maps and expectations concerning the equipment. 1.8 When all else fails, models are necessary when systems standardise become very large and complex. Some metaphors are successful in conveying the essential idea to the user, In an ideal world, the specific design Most mental maps and models are based which reduces mistakes and shortens the principles discussed above should help us on simple metaphors that should be easy time to learn a procedure. But metaphors to overcome most user-problems and to understand and remember. A metaphor may also be detrimental to the user, make all users perform perfectly all the is an intuitively understood analogy or increasing the number of mistakes made time. Unfortunately, since we do not live parallel from everyday life of a function and the time it takes to learn the process. in an ideal world, this is seldom the case. or process. Some metaphors just natu- Metaphors can be psychologically rally come to mind, while others are unsound and counterproductive in some When arbitrary mapping, arbitrary codes specifically created to help the user applications (i.e. the bin or waste-paper and unnatural operating procedures can- understand and use a complex process or basket metaphor of Macintosh to eject not be avoided, the only way may be to product. diskettes – the last thing you want to do standardise the operating procedures, is to throw your newly made back-up instructions, controls, signals, messages, copy in the bin). prompts, etc. to ascertain that users will

92 Telektronikk 3/4.1997 be able to learn and use the equipment and the services it provides efficiently. By having universal standards we may be able to learn and use inconsistent and incomprehensible procedures.

1.9 To sum up

All the ideas discussed so far can best be summed up in the two statements made by Donald Norman [2]: 1. Make sure that the user can figure out what to do. 2. Make sure that the user can tell what is going on.

If these two basic recommendations are always followed, it is difficult not to design a workable user-interface.

2 Basic communication abilities

Like ordinary telephony, videotelephony is supposed to make use of our natural communication abilities – i.e. the face-to- face conversation. It is presumed that we are all able to communicate fluently in Figure 4 The Tandberg videotelephone developed in co-operation with Norwegian normal face-to-face situations, and most Telecom Research (now Telenor R&D) in 1991 people certainly do master this form of communication excellently. But it is not necessarily true that we are fluent communicators in all situations. When people must conduct a face-to-face conversation over a videotelephone, unexpected problems arise because the videotelephone actually changes the way we carry on a conversation, besides posing problems of situations where they may actually be 2.1 Eye contact vs. face contact mastering the technology (i.e. call set-up detrimental to communication. and adjusting terminal functions). These The issue of eye contact in videotele- problems warrant a closer study. It took over a hundred years to learn to phony has been hotly debated. It is gen- use the ordinary telephone and there is a erally believed that eye contact is just as In a live face-to-face conversation you continuing need for educating users as important in videotelephone communica- will nod your head and use your hands new technology and new services and tion as in natural face-to-face conversa- and other forms of body language to functions are added. Adding to this the tion. Some ingenious designs have been show that you are listening and to indi- complication of transmitting and receiv- proposed to overcome the problem of cate your approval or disapproval. In the ing a live picture will not make things parallax error resulting from the need to sound telephone you have to resort to easier for the user of the videotelephone. mount the video camera outside the other strategies to keep the conversation viewing screen. flowing, such as ‘acoustic nodding’ or When designing the user-interface for the ‘audio prompting’, i.e. substituting videotelephone, some basic design goals However, recent findings do seem to sounds for visual signs to show that you were set up to guide us in our task: The indicate that face contact may be suf- are still on the line and paying attention. videotelephone had to be easy to under- ficient for obtaining proper communica- stand and easy to use, the procedures had tion between two people over the When speaking on the videotelephone, to be easy to learn by first-time users and videotelephone [3]. In videotelephones we tend to continue using the acquired easy to remember by infrequent users. with a low bit rate the diminutive screen audio telephone habits, even though we size and the reduced screen resolution should revert to our natural conversation may preclude proper eye contact in any strategies – the telephone habits are hard case, thus rendering any technical solu- to get rid of and are perpetuated in new tion to avoid parallax unnecessary.

Telektronikk 3/4.1997 93 2.6 Grouping according to function

Keys and controls should be logically grouped according to their function. The groups should be easy to discern and recognise, both visually and tactile. Groups should preferably have keys in contrasting colours or different sizes to make them more conspicuous.

2.7 Simplicity and consistency

Simplicity and consistency must be maintained. The number of keys and controls should be kept to the absolute minimum consistent with the basic functions of the videotelephone. Too many new keys, labelled with new function names or strange symbols, are certain to confuse most users. On the other hand, it is not advisable to use so few keys that each key must perform several functions.

Figure 5 Another Tandberg state of the art videotelephone with detached display screen and camera. 2.8 Context independence The Tandberg Vision 770 Keys and controls should not be context dependent, i.e. they should not have one function when the terminal is in one mode and a different function when the terminal is in another mode. Keys that change function according to context will The important design rule is to keep the will be in parallel use for many years) confuse most people and increase mis- camera lens as close as possible to the and it will avoid confusion and errors, takes. An exception to this rule are the edge of the screen. Most viewers prefer a especially for visually disabled people. number dialling keys, as they may be picture where the camera looks slightly used both for dialling and for entering downward from above the screen, rather 2.4 Ease of learning number codes and digits in other modes. than looking slightly upwards from below the screen. Rapid learning of the end-user proce- 2.9 Initiating a call dures for basic videotelephony is an – ‘off-hook’ 2.2 Compliance with ordinary essential design goal. The aim is to make audio telephony a videotelephone call as similar as pos- Since videotelephones must also function sible to an ordinary telephone call. The as ordinary ISDN telephones, videotele- The basic user-procedures in ISDN user-procedures in videotelephony phones must have a handset. On the ordi- videotelephony should comply with should support the conventional face-to- nary telephone, a call is initiated by lift- those adopted for ISDN telephony and face conversation between two people, ing the handset off-hook. The videotele- conform as closely as possible to the well without detracting their attention with phone handset must retain all the func- established user-procedures in ordinary complicated control procedures and tions that it has in ordinary telephones, audio telephony. Any changes made to status indications. i.e. going into dialling mode when the the traditional user-procedures should handset is lifted off-hook and terminating only be allowed where no other practical 2.5 Left-to-right control layout the call when replaced on-hook. alternative is possible (e.g. en-bloc dialling, hands-free off-hook/on-hook, The layout of keys and controls should When the videotelephone is in a hands- video enable/disable, etc.). be organised in a left-to-right progres- free or loudspeaking mode, it must be sion, i.e. keys and controls that are used possible to initiate the terminal for a call, 2.3 Number keypad first to initiate a call should be placed i.e. to go off-hook, without first having to furthest to the left, and those that are lift the handset. Thus, a separate control The layout of the videotelephone dialling used later, during a call, should be placed or off-hook key will be required to initi- keypad should conform to the standard further to the right. It may not always be ate hands-free or loudspeaking video- layout specified in the ITU-T Recommen- possible to do this since some controls telephone calls. dation E.161. This will ensure con- are used both before and during a call sistency with ordinary telephony (which (e.g. self-view).

94 Telektronikk 3/4.1997 2.10 En-bloc dialling

In those ISDN services where en-bloc sending of the telephone number is used, an extra function control, i.e. a delimiter key for ‘sending’ the number, is required.

Experimental evidence from Norwegian Telecom [4] highlights the high design sensitivity of this solution and the importance of proper user-guidance in the dialogue. Various visual and acoustic prompts may be essential to inform the user that the delimiter (i.e. the ‘send’) key must be activated after a telephone number has been entered to actually transmit the number to the exchange. With the widespread use of GSM mobile telephones this concept of en-bloc dialling is fast becoming common knowledge and the use of the ‘send’ number/ ‘off-hook’ key and the hook-on key is fast gaining acceptance.

2.11 Self-view

A self-view function, i.e. to be able to see yourself on the screen, is mandatory in Figure 6 Mock-up of a design exercise for a flat screen videotelephone from a student at the videotelephony. Self-view is necessary School of Industrial Design, Oslo for positioning yourself correctly in front of the camera. When self-view is in use it should not stop transmission of the outgoing video picture, since this may disrupt communication.

If the picture is displayed the same way and/or acoustic indications are provided in self-view as it is transmitted the user 2.12 Visual reciprocity in face- to advise both the caller and the called will have problems because movements to-face communication party of the video call. For the benefit of will be ‘reversed’ (if you move left, the blind and visually disabled people, spe- image on the screen moves right, and In face-to-face videotelephone calls cial acoustic indications are necessary. vice versa). Ideally, the self-view picture visual reciprocity should apply as the should therefore be reversed left-to-right, default mode, i.e. whenever A can see B, like the image we see of ourselves in a B should be able to see A. However, dur- 2.13 Video picture enable/ dis- mirror, to facilitate getting into the right ing a call objects and pictures may be able position in front of the camera. When transmitted from a subsidiary video cam- documents or objects are shown in self- era by either party without inhibiting the An important design goal is that there view by the internal camera on the screen picture received from the other party. must be no possibility for either party to of the videotelephone, a mirror image inadvertently send video and thus, un- will make it difficult to e.g. read printed Also, when one party makes use of the willingly, being seen by the other party. material on the screen, but a mirror self view function, the video picture in image will still make it easier to position the other direction should still be trans- To ensure protection of privacy, either the material properly before the camera. mitted normally, even though one party party must be able to stop outgoing cannot see the other party when self-view video, while keeping the audio channel The self view key should be a latching is active. If full visual reciprocity must open, by using a special control key, key (i.e. push once to activate, push once be adhered to at all times, it may seri- either during call set-up or during the more to turn the function off) and it ously disrupt the natural flow of commu- call. When a key is used it should have a should have a visual indicator to show nication and make several intended uses latching action and a clear visual indica- when the function is on. of the videotelephone impractical or tion showing whether the function is impossible. active or not.

Videotelephony is potentially more intrusive than audio telephony. To protect privacy, it is important that clear visual

Telektronikk 3/4.1997 95 In handset mode, the call is terminated References by putting the handset back ‘on-hook’, as in ordinary telephony. In a loudspeaking 1 Tetzner, K. Das erste Fernsehtelefon hands-free mode, the call will have to be der Welt. Funkschau, 58, 42–44, terminated by activating a special termi- 1986. nation-of-call or ‘on-hook’ key. 2 Norman, D A. The psychology of Since it is much easier to forget to oper- everyday things. New York, Basic ate an ‘on-hook’ key than to put a hand- Books, 1988. set back ‘on-hook’, a visual prompt may be needed to alert the user that the call 3 Kellner, B et al. Bildtelefon mit has not been disconnected. Blickkontakt? – Technische Möglichkeiten und empirischen Untersuchungen. NTZ, 38, 698–703, 3 Videotelephony – the 1985. new communication 4 Hamnes, K, Nordby, K, Anderson, form D. Testbed for user-system interaction. Paper presented at 13th Interna- Videotelephony is different in many tional Symposium on Human Factors ways from ordinary sound telephony. in Telecommunications, Turin, Ordinary telephony has been with us for September 1990. more than a hundred years, but there are still people who feel uncomfortable when Anderson, D M, Hamnes, K. A note using the telephone; such skills have to on block sending. Reporting on Figure 7 The use of a videotelephone is not always be learned and developed. results of pilot tests. ETSI TC HF 1 appropriate. Issues of privacy are important in video- (temporary document), 1990. telephone design Hopefully, it may not take another hundred years before we all will master Colins, S C, Rabbitt, P M A. Age and videotelephony, but it is a new mode of design. Age & Cognitive Perfor- interpersonal communication that has to mance Research Centre, University be learned. As we get used to the of Manchester, 1990. videotelephone we will develop skills in using this new medium, and in due Galer, I (ed.). Applied ergonomics 2.14 Terminating a call course we may become as proficient in handbook. Second edition. London, – ‘on-hook’ its use as we are today with the sound Butterworths, 1987. telephone. All our experience so far has It is important that a well defined and clearly shown that there are great poten- Maddix, F. Human-computer inter- secure termination-of-call procedure is tials in video communication and that the action : theory and practice. New provided, both to avoid unnecessary procedures necessary to use a videotele- York, Ellis Horwood, 1990. expenses and to protect against intrusion phone can be quickly learned without too on privacy. much trouble once its usefulness in vari- Shneiderman, B. Designing the user- ous applications is appreciated. interface. Strategies for effective human-computer interaction. New York, Addison-Wesley, 1987.

Knut Nordby is Senior Research Psychologist at Telenor R&D, Kjeller, where he has worked since 1985. He is Chairman of the ETSI Technical Committee Human Factors (TC HF), of which he was a founding member. He is member of the Permanent Steering Committee of Human Factors in Telecommunications (HFT) and Adjunct Professor teaching Man-Machine Interaction at UNIK, Center for Technology at Kjeller, University of Oslo. e-mail: [email protected]

96 Telektronikk 3/4.1997 A tactile marker for telephone cards: – From idea to global standard KNUT NORDBY

1 Background park cards, security lock cards), public card-reading device or pulled in a partic- transport (season tickets, taxi-cards, ular manner through a swipe-card reader. ‘Smart’-cards, i.e. machine readable ID-1 commuter passes), customer loyalty type cards bearing an IC (integrated cir- cards (petrol cards, frequent flyers cards, This ubiquitous ‘plastic’ card was first cuit) chip or a magnetic stripe, are being shopping benefit cards), medicine and introduced in the 60s when a number of used in an increasing number of societal social services (‘medicards’, social secu- restaurants in New York agreed on a applications, both in the professional and rity cards, ID cards), leisure interests credit service for selected customers leisure spheres. Applications cover such (library cards, ski slope or swimming whereby these could take guests to lunch diverse areas as finance (bank cards, pool day-pass cards, club membership or dinner and be billed later. As proof of credit cards, debit cards, ‘electronic cards) and, not least, telecommunications membership, each customer was issued a purse’ cards), access control and security (prepaid telephone cards and SIM cards plastic membership card with embossed (admission cards, hotel key-cards, car- for GSM telephones) – the list can be name and membership number. From extended indefinitely (see Figure 1). this card a carbon copy impression could When used, most of these cards must be be made with a roller on the receipt to be inserted in a specific orientation into the signed by the cardholder. This system proved so successful that many more restaurants, hotels and various shops joined the system – the credit card had been born – and today hundreds of millions of cardholders can use their cards with millions and millions of vendors.

Figure 1 Machine readable cards have invaded a large number of professional and leisure activities in society, and new applications are added each day. However, many people experience problems in orientating and inserting ID-1 cards correctly in the card readers, irrespective if these are electronic door locks in hotels or automatic banking machines, often making several attempts before finding the right orientation

Telektronikk 3/4.1997 97 RUNE LISLERUD/SAMFOTO Front

integrated circuit 53.98mm

Back 0.76mm thick

magnetic stripe

R 3.18mm

85.60mm

Figure 2 The most common and frustrating problem for Figure 3 The measures of ID-1 machine-readable cards all users of mini-banks (automatic tellers to some) and according to ISO standards 7810 and 7811 similar technology, is to insert the card in the correct orientation. By putting a tactile marker on the card, it will be much easier to turn the card to the proper orientation before inserting it in the card-reader. This will not only make the service both simpler and friendlier to users, but may be essential to blind and visually impaired users the modern ‘smart’-card, which has also bit small to be handled comfortably by been standardised (ISO, 1985). adult human hands, but normally this is not done very often. This ‘smart’-card technology was soon adopted by the telephone operators and is As smart-cards are becoming more and The physical dimensions of the card (see now used mainly for the common pre- more common, it is essential that the Figure 3) were inherited from the em- paid telephone-card and the SIM (Sub- cards can be utilised by all cardholders. bossed metal printing plates used in scriber Identification Module) card for One important prerequisite for using an address printing machines (e.g. Address- GSM telephones. In the telephone card, a ID-1 card is that it can be orientated cor- O-Graph) and have now been standard- specific number of tariff units have been rectly for insertion into the card-reader. ised (ISO, 1985) (see Figure 3). electronically stored on the card. When People who cannot orientate such cards the card is used, these units are deducted correctly by visual cues alone, e.g. blind By applying a magnetic stripe to the one by one until the card expires. and visually impaired people, may thus upper rear side of the card (above the be excluded from using important or embossing), cards were made machine- The SIM holds information in the IC chip even vital societal services (e.g. with- readable, and soon the card was taken up to authenticate a subscriber to use a par- drawing cash, buying the daily neces- as a personal identification (ID) card by ticular GSM subscription. Initially, the sities, using public transport, or making banks and in a large number of other SIM card was the same size as the ordi- telephone calls). security applications. When a pro- nary ID-1 type telephone card, but, as To aid blind and visually impaired users grammable, electronic, integrated circuit mobile phones grew smaller and smaller, to insert ID-1 type telephone cards chip was incorporated into the front left- the SIM was too large to fit inside. A correctly in the card-reader, the cards hand side of the card, it became possible small piece of the SIM card, bearing the must be physically marked in such a way to read and write large amounts of infor- IC chip with contacts, can now be broken that their orientation can be determined mation on the card, paving the way for off the card and installed in the GSM unambiguously by the sense of touch. telephone. This plug-in SIM module is a

98 Telektronikk 3/4.1997 Although such a tactile marker is essential for people who cannot rely on visual cues, people with no visual impairment may also experience problems with card orientation under adverse conditions, e.g. low illumination or inclement weather, and should benefit from a tactile marker to orientate cards. In situations involving safety (e.g. using a card while driving or while performing a hazardous operation) vision should not be used to orientate the card. Although the sense of touch is utilised in a large number of daily activities, e.g. identifying one key among several on a key-ring or finding the right coin among several in the pocket, tactile information is clearly an under-used information mode, compared to the visual or auditory modes.

There are at least six different tactile markers currently in use on ID-1 type telephone cards around the world (see Figure 4), but most telephone cards (like all other ID-1 type cards) do not have any tactile marker. The time was thus long overdue for creating an international standard for a common tactile marker on all ID-1 type telephone cards that must be presented in a specific orientation to the card reader. European Telecommuni- cation Standards Institute (ETSI) Tech- nical Committee Human Factors (TC HF) started work in 1994 with the aim of approving a common harmonised standard for a tactile marker for ID-1 type telephone cards. Thin flexible (TFC-1) cards, which have the same dimensions as ID-1 cards (e.g. like the telephone cards currently used in Italy or the tickets on the London Underground) are not covered by this standardisation, but the same general principles should also apply to these cards.

Liaison was established with the Euro- pean Committee for Standardisation (CEN) TC224/WG6 to agree on a com- Figure 4 Examples of various tactile markers on pre-paid telephone cards that are cur- mon standard for all ID-1 type cards to rently in use or have been in use in Europe. The broken off corner on Italian thin flexible avoid different standards for telecommu- (TFC-1) telephone cards is strictly not a tactile marker but is the result of authorising the nication cards and banking cards. In use of the card to prevent fraud 1992, work was also started in CCITT, Study Group 1, Question 18 (now Inter- national Tele Union (ITU-T) Study Group 2, Question 17), with the aim to adopt a global recommendation identical to the ETSI standard tactile marker for ID-1 type telephone cards. carried out by Norwegian Telecom Re- purpose. In these tests the subjects’ task search (now Telenor R&D) to determine: was to sort stacks of ID-1 cards, with and 2 User tests 1) Whether or not a tactile marker actu- without tactile markers, to the same ally will aid users to orientate telephone orientation. As part of the standardisation work and cards correctly, and, 2) which of seven to provide empirical data, user tests were tested tactile markers was best for this

Telektronikk 3/4.1997 99 65 tellerskritt 65 tellerskritt

Cut-off corner No tactile marker

Figure 5 The two ID-1 card types used in Test One

A total of 92 subjects participated in the The subjects’ task was to sort and re- ranking for each card category were two tests. They included people who stack the cards in each stack so that they computed. The results from Test One are were blind (N = 17), partially sighted were all oriented in the same direction. shown in Table 1. Paired samples T-tests (N = 33), elderly, mentally retarded and The task was to be completed as fast as between the mean time used to complete people with no apparent disability possible. When re-stacking the cards the task (t = 6.60, p < 0.01), the mean (N = 42), among them some 16–17 year with tactile markers, the subjects were number of errors made (t = 5.79, old students. Some of the subjects had instructed to “stack the cards in the same p < 0.01) and the mean preference and some had not used ID-1 cards before, orientation so that the tactile markers ranking (1.06) for the card with the cut- but none of the subjects had previously were on top of each other”. When re- off corner clearly show that a tactile used cards with tactile markers. stacking the cards with no tactile marker, marker significantly aids all users in the subjects were instructed to “place the orientating ID-1 cards quicker and with 2.1 Test One cards ‘picture-side-up’ with the text on fewer mistakes. each card oriented in the same direction”. The subjects were also asked to give a Of the 17 gravely visually impaired sub- In this test the subjects (N = 75) were subjective preference ranking of the two jects who participated in the test, 15 presented with two stacks of 20 random- card types (1 for best and 2 for worst). could not complete the task of re-stack- ly oriented, graphically marked, tele- ing the visually marked cards in the con- phone cards. The cards in the experi- The time required to perform the task, trol stack, since they obviously could not mental stack had a tactile marker in the the number of errors made in each task make use of the visual markings on the form of a cut-off corner (see Figure 5, left and the card preference rankings were cards. Their data are not included in the panel), while the cards in the control recorded. Mean time used, mean number results. However, they could all easily stack had no tactile marker (see Figure 5, of errors made and mean preferences execute the task of re-stacking the cards right panel). with a tactile marker in the experimental stack. For blind users, a tactile marker is essential for orientating ID-1 type cards.

2.2 Test Two Table 1 Paired samples T-test, Test One (N = 75) In Test Two the subjects (N = 92) were UNMARKEDERROR VS. MARKEDERROR UNMARKEDTIME VS. MARKEDTIME presented with seven different stacks of 20 randomly oriented, white ID-1 type H = UNMARKED = MARKED H = UNMARKED = MARKED 0 ERROR ERROR 0 TIME TIME cards. Each of the stacks had a different H1 = UNMARKEDERROR > MARKEDERROR H1 = UNMARKEDTIME > MARKEDTIME tactile marker: 1) Small diameter notch; 2) 2 mm notch; 3) Embossed arrow MEAN DIFFERENCE = 1.667 MEAN DIFFERENCE = 19.107 s. (Braille dots); 4) 0.7 mm notch; 5) Hole- SD DIFFERENCE = 2.495 SD DIFFERENCE = 25.059 in-card; 6) Cut-off corner, and 7) 0.3 mm notch (see Figure 6 for details). T = 5.785, DF = 74, p < 0.01 T = 6.603, DF = 74, p < 0.01

100 Telektronikk 3/4.1997 Small diameter notch 2 mm deep notch

Embossed arrow 0,7 mm deep notch

Hole in card Cut-off corner

0,3 mm deep notch

Figure 6 The seven tactile markers used in Test Two

Telektronikk 3/4.1997 101 Table 2 The results from Test Two (N = 92) 2.2.1 Cut-off corner The cut-off corner card required the least 0.7 mm Cut-off 0.3 mm 2.0 mm Hole in Emboss Small time to orientate. In terms of errors, it notch corner notch notch card arrow notch came only fourth, but it should be noted that one subject was responsible for 11 of N 77 92 92 91 92 92 89 a total of 16 errors made and that the dif- Mean error 0.72 0.17 1.62 0.08 0.06 0.13 0.55 ferences in number of errors between the four best markers is small compared to N 79 92 92 92 92 92 91 the number of errors made in the worst cases. The cut-off corner card was ranked Mean time 64.86 44.41 66.33 47.15 47.55 55.26 49.36 first by most subjects.

Overall, the cut-off corner was judged best in Test Two. For technical reasons, however, it cannot be used on ID-1 tele- The subjects’ task was to orientate and orientate the cards with the cut-off corner phone cards (it corrupts the magnetic re-stack the cards in each stack so that in Test Two is nearly identical to the time stripe and makes card feeding difficult). they were all oriented in the same direc- obtained for cards with this tactile It is currently in use on thin flexible tion. The task was to be completed as fast marker in Test One, thus confirming the (TFC-1 type) telephone cards in Italy. as possible. When re-stacking the cards, inter-test reliability. The results obtained the subjects were instructed to “stack the with the 0.7 mm notch and the 0.3 mm cards in the same orientation so that the notch are similar to or worse than the 2.2.2 2 mm notch tactile markers were on top of each results obtained with the unmarked cards The 2 mm notch was second in terms of other”. The subjects were also asked to in Test One. Thus, no real benefit would time used, second in terms of errors give a subjective preference ranking of be gained by using these two tactile made and second in the preferences each tactile marker (1 for best and down markers, except for young blind people ranking. Overall, the 2 mm notch was to 7 for worst). with a good sense of touch who are judged second in Test Two. There are no totally dependent on any form of tactile valid technical objections against its use The time to perform the task, the number marker. It should be noted that most on ID-1 type telephone cards. This tactile of errors made and the tactile marker blind people are elderly who have re- marker is currently in use in Denmark. preference ranking were recorded for duced skin sensitivity, and most people each card stack. Mean time used, mean who are blind in the western industrial- number of errors made and mean prefer- ised societies (Europe and North Amer- 2.2.3 Hole in card ence ranking for each tactile marker was ica) become blind as the result of Cards with a hole came third in terms of computed. To minimise any position or untreated Diabetes mellitus, which often time used, first in terms of errors made learning effects, Test One and Test Two results in loss of vision through retinopa- and third in the user preference ranking. were randomly assigned, and within tests thy, a vascular disorder of the retina. But Overall, the hole in card was judged to the presentation order of the different diabetes also destroys the capillaries of be third in Test Two. For technical card stacks was randomised. the skin, thus reducing the sense of reasons, however, this tactile marker can- touch. Although some tactile markers not be recommended since the hole cor- The results from Test Two show clear came better than others in Test Two, not rupts the magnetic stripe area and any differences between the seven tactile all can be recommended for various tech- future optical storage area. markers (see Tables 2 and 3). It should nical reasons. be noted that the mean time required to 2.2.4 Small diameter notch The small diameter notch was fourth in terms of time used, fifth in terms of Table 3 Time, error and rankings in Test Two errors made and fourth in the user preference ranking. Overall, the small diameter Overall ranking All subjects (N = 92) Blind subjects (N = 50) notch was judged to be fourth in Test Two. There are no technical objections Tactile marker Time Errors Rank Time Errors Rank against its use on ID-1 cards. This tactile marker was used on telephone cards in 1. Cut-off corner 141141 Norway during a short trial period. 2. 2.0 mm notch 222322 3. Hole in card 313233 2.2.5 Embossed arrow Cards with an embossed arrow in raised 4. Small dia. notch 454465 dots came fifth in terms of time used, 5. Embossed arrow 535514 third in terms of errors made and in the user preference ranking. Overall, the 6. 0.7 mm notch 667657 embossed arrow was judged to have 7. 0.3 mm notch 776776 come fifth in Test Two. However, this

102 Telektronikk 3/4.1997 PER-ANDERS ROSENKVIST/SAMFOTO

Table 4 Overall rankings in Test Two

All subjects Blind & visually impaired (N = 92) subjects (N = 50)

1. Cut-off corner 1. Cut-off corner 2. 2.0 mm notch 2. 2.0 mm notch 3. Hole in card 3. Hole in card 4. Small diameter notch 4. Embossed arrow 5. Embossed arrow 5. Small diameter notch 6. 0.7 mm notch 6. 0.7 mm notch 7. 0.3 mm notch 7. 0.3 mm notch

tained with these subjects (N = 50) were singled out for separate analysis. The Figure 7 The prepaid telephone card is a most useful service both to the operators, results are practically the same as for the who do not have to collect the coins from the terminals or have them vandalised by whole group. Time and error rankings for people who are after the cash, and to the users who do not have to worry about carry- both groups are shown in Table 3 and ing the right coins. However, it is important that the users manage to insert the cards subjective ratings in Table 4. in the right orientation. Applying a tactile marker to cards simplifies insertion and makes it possible for blind and visually impaired people to utilise the service An interesting observation, which we have not seen reported before, was that blind people, when exploring the cards with their fingers, felt the edges (usually the long edges before the short) before they felt the surfaces of the cards. Thus, a tactile marker cannot be used as the 2.3 Blind and visually impaired tactile marker on the edge of the card increased thickness of the card prevents subjects should be detected before a marker on its insertion into card-readers and the the surface of the card. This also applies magnetic stripe area and any future opti- Of the 92 subjects who participated in the to people with normal or impaired vision cal storage area are corrupted. trials, 17 were completely blind and 33 when handling a card prior to insertion in were visually impaired. The results ob- a card-reader. 2.2.6 0.7 mm notch The 0.7 mm notch was sixth in terms of time used, sixth in terms of errors made and seventh in the preferences rating. Overall, the 0.7 mm notch was judged sixth in Test Two. This tactile marker was proposed by CEN TC 224/WG6. It should be noted that this marker did not do any better than the unmarked cards in Test One.

2.2.7 0.3 mm notch The 0.3 mm notch was seventh both in terms of time used and errors made and sixth in the preferences rating. Overall, the 0.3 mm notch was judged seventh and last in Test Two. This tactile marker has been used on several telephone cards in Europe. This marker did not do any better than the unmarked cards in Test One. Figure 8 The first two telephone cards, from British Telecom and Telenor, to bear a tactile marker according to ETSI ES 300 767 and ITU-T Recommendation E.136

Telektronikk 3/4.1997 103 Front Cloke of BT Labs at Martelsham Heath, UK, to determine which of the two positions of the 2 mm notch was more acceptable. The two positions were the original 2 mm notch position on the lower long edge of the card and the 0.3 mm notch integrated circuit position on the right hand short end of 2mm the card. The test showed that the position on the long bottom edge was preferred to the position on the short right R 12mm hand end, but the diffrence between the two positions was small and not signifi- 15mm cant.

On the strength of this finding ETSI changed the position of the tactile marker in the next version of the Draft Standard, Back and was soon followed by ITU-T 17/2. CEN TC224/WG6, however, came up magnetic stripe with a proposal of a 0.7 mm deep notch on the long bottom edge of the card, i.e. the same position as the original 2 mm notch! This led to some confusion. How- 2mm ever, ETSI TC HF and ITU-T 17/2 could not hold up their standardisation activi- R 12mm ties any longer and the proposals for a 2 mm deep notch on the right hand short end of the card went through the formal 15mm adoption procedures (see Figure 9).

On Monday 26 May 1997, the ITU-T Study Group 2 plenary meeting in Figure 9 The measurements of the tactile marker for telephone and SIM cards Geneva unanimously adopted IUT-T according to ETSI ES 300 767 and ITU-T Recommendation E.136 Recommendation E.136 “Tactile Identi- fier on Pre-paid Telephone Cards” which thus became a global standard – thereby beating ETSI by a month.

On Friday 27 June 1997, ETSI Vote 97/26 ended with the result that the 2.4 Recommendations 3 The standards prETS “Human Factors; Telephone prepayment cards; Tactile identifier” was The results from both tests showed that a On the strength of the results of the user adopted as ES 300 767. As this goes to tactile marker on ID-1 type cards was trials, the 2 mm notch was proposed in press, BT has decided to adopt the ETSI/ preferred to not having a tactile marker, the first ETSI Draft Standard as a Euro- ITU standard tactile marker for their tele- and a tactile marker significantly aids pean regional standard for a tactile phone cards. Other operators are sure to card orientation for all users. The results marker for telephone and SIM cards. follow suit (Figure 8). from Test Two clearly show that the more Also ITU-T proposed this tactile marker prominent or salient the tactile marker is, in the first draft recommendation. How- In a belated proposal from CEN TC the better it is. Although the cut-off ever, the liaison with CEN TC224/WG6 224/WG6 on prEN 1332-2 “Identifica- corner marker came best in the trials, made it clear that the chosen position on tion Card Systems – Man-Machine Inter- slightly ahead of the 2 mm notch, it could the lower long edge was not acceptable face Part 2: Definition of a Tactile Identi- not be recommended for technical since it would interfere with the transport fier for ID-1 cards”, it was now proposed reasons. Thus, the 2 mm notch was the of cards in some automatic banking to cut off the lower right corner radius clear recommendation, both on usability machines. Also, CEN wanted a much (which previously had been ruled out as and technical grounds. The results from shallower notch than 2 mm (initially they totally unacceptable) as an alternative to the full group and the results from the opted for 0.3 mm, but eventually the 2 mm deep edge indentation. What- sub-group of blind and visually impaired accepted 0.7 mm). Our trials had shown ever kind of tactile marker is adopted, subjects are virtually the same (see that the 0.3 and 0.7 mm notches were no some party may find objections. Tables 3 and 4) and our recommendation better than unmarked cards (with the is the same for both groups. The two exception of young blind people without The main objection against a tactile tactile markers that came last in the trials any other impairment). The position of marker has been cost. The cost of app- (0.7 mm notch and 0.3 mm notch) were the notch, however, was open to discus- lying a tactile marker is negligible com- no better than unmarked cards. sion. A test was carried out by Mr. T. pared to the cost of the whole card. The

104 Telektronikk 3/4.1997 ETSI ETR 165. Human Factors; Recom- mendation for a Tactile Identifier on Machine Readable Cards For Telecom- munication Terminals. Januar, 1995.

ETSI ES 300 767. Human Factors; Tele- phone Prepayment Cards; Tactile Identi- fier. 1997.

Figure 10 Promotion telephone card issued for the 16th International Symposium on Human Factors in Telecommunications, Oslo, Norway, 12–16 May, 1997 (HFT’97)

main objection against the new standards ISO 7811/4. Identification cards Part 4: is that some card manufacturing equip- Location of read-only magnetic tracks. ment for transporting the cards from 1985. station to station may have to be modified, but compared to the expected ISO 7811/5. Identification cards Part 5: growth in revenue from increased card Location of read-write magnetic track. usage these costs should be minimal, 1985. especially since such equipment must be replaced regularly as it is worn out. ISO 7816-1. Identification cards, Inte- grated circuit(s) cards with contacts Part The important lesson here is that user 1: Physical characteristics. 1987. trials are essential to determine the optimum form and location of a tactile ISO 7816-2. Identification cards, Inte- marker on telephone cards to make the grated circuit(s) cards with contacts Part cards easier to orientate for all users. 2: Dimensions and location of contacts. Although this may be in conflict with the 1988. interests of some industrial parties, the interests of the users should be given ISO DIS 11694. Identification cards, priority – in the end it is the customers Optical memory cards Part 2: Dimen- who will determine if the telephone cards sions and location of the accessible will be used or not. optical area. 1995.

References

ISO 7810. Identification cards : Physical characteristics. 1985.

ISO 7811/1. Identification cards Part 1: Knut Nordby is Senior Research Psychologist at Embossing. 1985. Telenor R&D, Kjeller, where he has worked since 1985. He is Chairman of the ETSI Technical Committee ISO 7811/2. Identification cards Part 2: Human Factors (TC HF), of which he was a founding Magnetic stripe. 1985. member. He is member of the Permanent Steering Committee of Human Factors in Telecommunications ISO 7811/3. Identification cards Part 3: (HFT) and Adjunct Professor teaching Man-Machine Interaction at UNIK, Center for Technology at Kjeller, Location of embossed characters University of Oslo. oncards. 1985. e-mail: [email protected]

Telektronikk 3/4.1997 105 The User-Interface Design process; some new year resolutions SIMON CLATWORTHY

This article looks at the interface technological complexity, new design However, our understanding of the design process and reflects upon its disciplines, an increased focus on good design process is becoming clearer, par- practice in projects. Methods and interface design, and better user inter- ticularly upon the focus of each of the disciplines are discussed based upon faces to compete against. It seems that different project stages and how each can recent project experience, and some finally, the Human Factors profession is be influenced. If I take a bird’s eye view future directions for interface design in the spotlight with expectations greater of the design process, then it has essen- are suggested. The article focuses on than ever before that we can produce tially three major phases which are criti- systems aimed at the consumer miracles. If Andy Warhol was right cal to the user interface design: a user market, although the distinction be- (everyone will be famous for 15 focus, a technological focus, and back tween home users and office users is minutes), our 15 minutes of fame has again. now becoming less and less distinct. started ticking, and I am afraid that the The paper takes a pragmatic and per- first 10 minutes passed without us The following view of the design process sonal view of the design process, rather realising it. We should therefore think is based upon recent project experience than an idealised view. This view is of carefully about how we use the remain- where human factors has had a reason- course my own, and does not neces- ing minutes. ably large role. I still notice, however, sarily reflect the views of the organisa- that many projects have a tendency to tions or projects I work for. skip the first stage and jump directly to The design process as she technical development. is practised Introduction The first phase: user focused The design process is becoming more The job of designing a good user inter- and more complex. Time to market is The first phase of a project has very face just gets harder and harder. The rate being squeezed at one end, whilst at the much a user focus. It defines goals, user of change in user interface development other end, an increasing number of groups/market segments, carries out is increasing rapidly, even though the experts need to be involved in the pro- analyses of needs, talks to users and gen- modern desktop is not so modern any- cess and professionally co-ordinated. In erally creates a user requirement spec in more (it must be reaching its 25th birth- addition, expectations of quality, novelty one form or another. It also often ends up day soon). The reason that the design and ‘sexiness’ in interface design are with some form of interface visualisa- process has become more complex is a increasing. tion; a rapid prototype. At this stage the combination of factors such as increased project is very much based upon a user model of a product or application with very little focus on implementation aspects. I call this phase the ‘white knight’ phase, it would like to be focused upon idealistic solutions and based purely upon user needs. Technical expertise has a back seat during this phase, increasingly too much so (“Yes we are sure it is important, but it comes later – we must not let it disturb the identification of real needs”). This phase usually takes quite some time and for good reason; accurately identifying user requirements and defining project goals takes time. We should respect this, but we somehow try to apologise and rush things through. Building a good project team takes time and cannot be rushed.

The second phase: technical implementation

The second phase of a project is the implementation phase, and has a clearly (often exclusively) technical focus. At this stage, the users, the marketing people and the interface designers usually have a back seat, and develop- This Internet page from Mercedes shows a good combination of design and user inter- ment moves quickly and seemingly effi- face skills. The design builds well upon Mercedes market profile, yet is very easy to ciently. This stage of the design process understand and navigate through. Although there are many menu alternatives on this is what I call the ‘black box’ phase. It page, the screen does not seem cluttered due to grouping of functions and typography. takes user focused input, translates it and A good example of a second generation web-page breaks it down into technical sub-compo-

106 Telektronikk 3/4.1997 nents and develops solutions in parallel. I really like the HUFIT tools (slightly Resolution #2: The parts are then put together and tested reworked as USERFIT) in this respect. If Avoid virtual ergonomic teapots to form a whole. The ‘black box’ phase you can get over the masses of forms of system development is resistant to which HUFIT requires, or even better, This is a trap I have fallen into loads of user influence, and sometimes directly tailor your own, then there is a lot to be times, and which I still have a tendency antagonistic towards it. At this stage, the said for their approach of forcing a pro- to fall into. It is based on the (mistaken) project is very much based upon a techni- ject to focus upon the discussion of the belief that poor user interfaces are the cal model of a product or application. implications of user requirements in a results of ‘nerds’ doing the designing, project. Forcing yourself to think about and that if only we (the user interface The final phase: user focused these things together with the stake profession) could manage (read: control) holders in a project (including the imple- the process, then everything would be mentation team) and then incorporating perfect. It is just not true, although I wish The final stage of a project is the com- them in the project mandate are some of it were. pletion, marketing and launch. This the great things HUFIT has taught me phase returns to a user focus, and many (thanks, HUSAT). Thanks also to Ken During the 1960s, the human factors proof the user disciplines from the first Eason and Susan Harker for their simple fession set about the design of the phase come back into the project. It is not yet effective task analysis method ‘an ergonomic teapot. Teapots are not partic- unusual that users and experts alike open systems approach to task analysis’. ularly user friendly. The point of balance receive a few surprises at this stage and is wrong, you can scald yourself, they find that the end result deviates from the intended result in a few major ways, and many minor ways. This stage I call the ‘amphetamine’ phase. Tight deadlines, technical problems and a product slightly wide of the market require everyone to work around the clock, on borrowed time. The final stage is not always a nice period in a project, which probably explains people’s unwillingness to document it properly, and to carry out the all- important de-briefing afterwards.

This is very much a simplification and a generalisation, but it highlights one thing which I think we are not very good at; managing the shift of emphasis between a user focus and technical focus in projects. The fact that these stages are so clearly visible in projects is actually in itself a sign that something is wrong. What should we do about it? Well, here are my new year’s resolutions for next year.

Resolution #1: Get more out of the early stages

I have noticed during the last year an increasing interest for a user focus at the early stages of a project, and an growing willingness to devote time to this area. However, projects increasingly lack a means of effectively documenting this, and even more importantly, extracting the relevant information from this phase. It is not unusual to see a user based task description in a project, yet find that nobody really has analysed this and said: “So what are the consequences for the project and for the user interface?”. User requirement experts need to become much better in pulling out the implications of our analyses and communicating The user interface design process is still too much based upon the relay race metaphor. them effectively. We should start playing football instead

Telektronikk 3/4.1997 107 We should, however, be aware of the consequences of moving towards a football team approach. In a relay, only one person is working at a time, although that person is the best in the field. You can also replace one or all of the relay team up until or even during the race. In football, at times all eleven players are working together, and it is wise that all eleven have played together before the match starts. Additionally, substituting more than one or two players underway has a negative effect upon the result as a whole. Football is therefore much more costly, requires a team captain, and in addition a greater degree of strategy and co-ordination. There again, football players (at least the good ones) get paid a lot more, and receive much more media attention, so no prizes for guessing what kind of sports metaphor I am putting my effort into in future. Gautam Ghosh from ISI as in Norway has a nice exercise in user requirement specification; the specification of a hammer, and the implementation of the spec. Well, you Resolution #4: did say a metal tipped object which can be held in the hand and used to impart force to Communicating that ‘a spade is nails so that they can be driven into softer materials in a chosen direction ... a spade’ (or alternatively, a tool for soil excavation and transportation)

This resolution is just a recognition of the problem of transition from a user model to a technical model of an application. If we stick with the relay race analogy for drip, you can never find the tea bag, etc., opportunity to show what it really can the time being, the user centred ‘baton’ etc. The project therefore redesigned the do. Now it seems that we are increas- has a different shape from the technically teapot from a purely ergonomics per- ingly being offered the opportunity to focused ‘baton’. The user focused spective and resolved most of these user show what we can do, and not sur- ‘baton’ is based upon how the human problems. They did however also pro- prisingly are making a few mistakes as mind works and how people best can duce one of the ugliest things in the we learn. Hopefully, increased ex- carry out tasks. The technical ‘baton’ is world. Not only that, it was difficult to perience should temper our ambition. based upon how computers store, retrieve manufacture. and present information and how to Resolution #3: structure bits to make the best use of this. User interface designers, particularly the Play more football and run Until processors and databases work in human factors discipline (and I am one of fewer relays the same way as the human mind, there is them), seem to want to create two kinds going to have to be a transition from one of ‘virtual ergonomic teapots’: One of the problems with the three to the other, and back again. As with all 1. Interfaces which work well on paper, phases mentioned earlier is the transition square pegs which have to go into round but which when technically imple- stages from user focus to technical focus holes, bits go missing underway. The mented cause all sorts of problems and and back again. If you think of the design important thing is to try and make sure alienate users (‘thou shalt have a good process as a relay race, it seems that the that the bits which get shaved off are not interface, but your system will suffer baton does not get handed over properly of critical importance. Many systems as a consequence’). at the early stages, and that again, it does look good at the specification stage, but not get handed back properly towards the become a mess when implemented, 2. Interfaces that are easy to use but end. mainly because the wrong bits went which are incredibly boring (‘thou missing underway. This is a communica- shalt be efficient but not have fun’). In fact, the problem might be the idea tion problem in the creation of a require- that there is a hand over at all and the ments specification, and we (user inter- What can we do about the problem? process should not be likened to a relay, face experts) need to be much better at Well, I have a feeling that it might go but to a football match, more of a team clearly communicating the critical parts away on its own after a while. The effort. This may be the direction for of a user requirements spec for projects, human factors/user interface profession future development, but at the moment those which should be cast in concrete lacks self esteem; it has always battled to we are still in the world of the relay race. and the less critical, those which can be be accepted as an important aspect of a adapted. product, and is just dying to have the

108 Telektronikk 3/4.1997 We can learn a lot from the oil and inter- guaranteed, and the development process towards computing at home, we are national standards world here. They have has an increased user focus as a whole. becoming aware of the need for a social a very clear notation for writing require- and cultural foundation for user requirements, based upon the two magical Ideally, we should hire in experts in user ments. I am sort of half jumping at the words shall and should. By being very requirements to form part of the project moment about this and resolve to jump clear about requirements which shall be team and then keep them in the team more next year to spread the net even implemented as opposed to those which throughout the design process. They wider. So far, we have not succeeded in should be implemented, and receiving should be responsible for managing user including this factor adequately for the project approval for this, the move from requirements capture, yet use project computing profession to take it on board, a user focus to a technical focus can be members to carry out some of the data but it is coming. improved. collection. This requires the ability to manage their participation without I was promoting the use of HUFIT Resolution #5: wasting resources. If experts become too earlier, and believe that it has a lot to Stop lugging the piano and start peripheral, then they may lose touch with offer the design process. In some ways, a band project development (therefore weaken- the methods developed in MUSIC (origi- ing your football team). If you involve nal name: Measuring Usability In Con- them too much, you may waste re- text) for evaluation have taken some of A famous jazz musician, who at one time sources. the HUFIT philosophy further. In fact, played several instruments, but finally one of the most useful parts of MUSIC, focused upon the saxophone, once said In a recent project I tried to improve the context analysis, is something I have that “the best thing about just playing the communication with the implementation used during user requirements capture saxophone is that you don’t have to lug a team during the requirements capture and it is a great way of structuring work. piano around anymore”. I get the feeling stage of a project by including them in Now I am stumped, because I need to that user interface designers have been the requirements capture (they carried have some methods which can be used to playing loads of instruments and have out user interviews, wrote parts of the step further from MUSIC’s context not only lugged around the piano, but URS), and we held HUFIT workshops analysis to include a social and cultural also the double bass, saxophone and together to discuss results and critical focus. The problem is, I cannot find one drums. More and more specialisms are aspects of the user interface design. This method which is design-process friendly. needed to create good solutions, and was working well, and we did at one Tips, anyone? being a sole representative of user inter- stage create a common understanding of face design in a project is backbreaking. user and technical needs. However, I also It is about time that we took the conse- Resolution #8: discovered the vulnerability of such a quences of this and involved other Move from the information team to change. At one stage of the pro- experts in the process. I am happy to play society to the experience society ject, several members of the implementa- just the saxophone in a good band (as tion team were changed, and all of the long as I can play a few solos). Peter Cochrane from BT has a point, we careful work in creating a cohesive team seem to be moving from the information went down the pan. Not daunted by this, society to the experience society. His slo- Resolutions #6: I have vowed to continue this approach, gan “instant gratification anywhere, any- Combine DIY and experts for but will now improve team building and time” does seem to be a little over the top user requirement capture? cohesiveness. So, again, knowing how to do things correctly has become simpler; One of the questions always asked in a being able to carry them out, more com- project is how much one should hire in plex. experts during the project and how much Table 1 The table shows how, over time, the focus of user the project team can do themselves (DIY). Resolution #7: interface design has moved from supporting tasks to Widen the net supporting social structures. Each time we widen the net, Experts in requirements capture will be we invoke complexity quicker, more thorough and will identify During the early days of user interface more requirements than non-experts. design, the focus was very much based They will also be better at separating Period Theme Challenge upon the layout of menus and navigation important requirements from less import- within hierarchical systems. At this point ant (see #4, above). They do, however, 1980–1990 Menu-based systems Menu structures, we jumped up and down and said “com- have an overhead. They cost money (not Time delays, puting is all about tasks and improving always more than if you do it yourself, Navigation task performance”. The net was duly depending on hourly rates) and have to widened to include this, and task analysis 1990–1995 Windows and GUI Supporting tasks, communicate the results to the develop- became a focal point. It gradually organisational ment team. dawned upon us that no matter how well context we focused on the task, it was not DIY can lead to ineffective resource 1995–1997 Internet Social and enough; the organisational context and usage, the use of incorrect methods or the cultural context environmental context were important. incorrect interpretation of results. How- We jumped up and down about this and 1997+ Internet at home Family structures, ever, the results become part of the the net was ceremoniously spread wider computing as an implementation process, continuity is to include this. Now, as we move experience

Telektronikk 3/4.1997 109 but he does have a point and it is going to I am also unsure of how much user re- One of the dangers when working with have a huge effect upon user interface quirements knowledge you should trans- designers is that it is easy to be blinded design. fer to designers and how much total free- by the light of design, just because dom should you let them have. One of results are presented so well. There is “With the gap between competing the goals of using designers is both to something incredibly seductive about products narrowing in terms of pull in people from outside a problem designers’ renderings during the first performance and quality, the ex- area, and at the same time bring in crea- stages of the design process; you become perience that a product delivers is tivity at a critical part of the design pro- convinced that the solutions are good, rapidly becoming the key to offering cess. This is usually at the same stage as just because they are presented so well. distinctive value to the customer.” user requirements capture, and I have Resolution for next year: watch out for Harvard Business Review tried many variants over the years. things that look sexy, but which will not Designers function well with some basic work. Train yourself to spot the differ- What does this mean for interface user requirements, but they become ence between novelty and truly good design? Well, it means that users will be rapidly constrained when inundated with design. expecting more from their interface than detail. Work with designers is often an they do at the moment. The computer evolutionary process, so you do not have games industry has always had a focus to drown them with information at the start. Conclusion – Less on this aspect, so we can expect that some of the elements of computer games drowning, more waving; I have had good results from including will be coming to an interface near you, designers and the project team in the The need for Interactive and soon. In fact, Nolan Bushnell, found- HUFIT workshops covering activity ing guru of Atari®, is a strong voice in design management analyses, context analyses and user anal- this respect. His article, Relationships be- yses. It gives the designers a thorough There is a great deal of benefit to be tween fun and the computer business, understanding of the user issues, and gained from having one person follow makes interesting reading. includes a short written summary. After user aspects throughout the process, this, the designers have been able to particularly to link across the phases It also means the introduction of new carry out their first concept generation from user focus to technical focus. The disciplines to the design process, particu- phase whilst I have fleshed out a more increasing number of skills involved in larly those from the marketing and media thorough User Requirements Specifica- user interface design means that one per- world. Internet has shown us the import- tion in parallel. In this way, the designer son cannot and should not have expertise ance of capturing a user’s attention with- has worked based upon knowledge of in all of them. Instead, a project needs to in the first five seconds. Like it or not, users and tasks, but without the con- hire in expertise within the project, and this trend is going to accelerate, so jump straints of the requirement spec. This someone should be responsible for co- aboard and put your seat belts on. In fact, allows a creative process for the ordinating this. It is therefore natural that the traditional human factors input to a designer, with (hopefully) the user-task- a management task is created for inter- project will probably lose some of its rel- context information unconsciously steer- active systems design. evance and take a back seat to design and ing them in the right direction. When the media disciplines as the goal for comput- designers start work on the concept Design management is a well established ing moves from an efficiency directed detail, I include the URS as a specifica- discipline within the traditional design development towards an experience tion document. I get a feeling that this world, and the interactive design man- directed one. gives the best of both worlds; it allows ager should be modelled upon traditional the designers to be creative, whilst giving design management, with added ingre- Resolution #9: them enough context. If anyone has other dients to cover the human factors and Prepare my employer for the variations on this briefing procedure, technical implementation aspects. User consequences of resolution #8 please get in touch. interface design has become a design discipline and is no longer purely a human I have a real dilemma here: how ana- Resolution #10: factors discipline. We should take the lytical should the design process for user Wear shades consequences of this and start to learn ow interfaces be? Traditional Human Factors to work together with other disciplines. is analytical, whereas Design is a cre- After all, we all have the same goal. One of the fantastic things about working ative, almost artistic process. If I ask if with designers is not just their creativity the two can be combined then I always Who then should the design manager but the combination of creativity and receive a hearty ‘yes’ as a reply. When I resemble? The skill make-up should be strong communication skills. I think that ask in which proportions they should be something like is: Industrial Designers have an edge over combined, and which one should come Graphic Designers here, with Architects • Strong founding in methods for user first, I do not receive many answers, and coming third (architects do however have requirement capture, task analysis, I am unsure myself. Obviously, the strong analytical skills and a focus upon survey design, etc. answer will vary from project to project, structuring information). I always find but then who should decide that project • Detailed knowledge of the design pro- working with designers a refreshing A should be 30 % Human Factors and cess experience, although the experience 70 % Design, whilst project B has en- varies greatly. • Basic knowledge of the human cogni- tirely different proportions? tive process and perception • Marketing and branding expertise

110 Telektronikk 3/4.1997 • Industrial and Graphical design knowledge, particularly awareness of different design groups’ styles • Media and Multi-media knowledge • Knowledge of technical implementation • Project management skills.

The role of the interactive design manager is to lead the user side of a project, maybe even the whole project, although more likely part of a larger team. The area of responsibility for the design manager is to ensure that adequate user requirements analyses are carried out in a project and that the project has a mandate which is clearly anchored on the user and market side. The design manager should have responsibility for developing a prototype and testing this. Further, he/she should have responsibility for the user requirements specification, the style- guide, the storyboard and to follow these from start to finish, with user testing underway. In addition, he will have responsibility for the user documentation, and for organising necessary training.

As for how many resolutions I will manage to get to grips with, only time will tell. At any rate, I will go out of this year knowing a lot more about what I do not know. Year 2000 problems anyone?

Simon Clatworthy is Research Scientist at Telenor Research & Development, where he has been employed since October 1996. He has worked with command and control systems and systems for public use, and more recently, Internet on the TV. During this time he became interested in the design process, and how best to integrate human factors methods. e-mail: [email protected]

Telektronikk 3/4.1997 111

112 Telektronikk 3/4.1997

Special

113 114 Telektronikk 3.1997 First installations of WDM, optical ADM and optical in-line amplifiers on long-haul cable links in Norway STEINAR M. SVENDSEN

Indisputably, the first optical wavelength division multi- Cable route Oslo – Bergen plexing (WDM) systems to be established in Norway, appeared almost ten years ago on a handful of regional Table 1 details the amplifier section lengths. On two sections short-haul routes in Oslo. Those were the glorious days of attenuation difference between the two fibres being used PDH, one 565 Mbit/s channel was transmitted in each of the exceeds 2 dB, otherwise 0.3 dB or less. Attenuation coefficient two windows, 1310 and 1550 nm. The WDM systems of 1997 per section is 0.23 – 0.29 dB/km, however, one value reaches imply the introduction of a new era. The test project re- 0.32 dB/km (Section 1). ferred to in this paper has successfully demonstrated transmission of four 2.5 Gbit/s channels over 520 km fibre cable It should be noted that the 120 km section from Hønefoss to Gol without any need for regeneration, optical equalization or comprises cable which was installed during the 1985–88 period, dispersion compensation between the terminals. when only operation in the 1310 nm window was foreseen. The basic problem of this ‘old’ fibre type is high mode field diameter and low cut-off wavelength. The remaining part of the Introduction route is double-window standard single mode fibre (SSMF - G.652). Early 1997 Telenor AS was offered to perform FVO (Field Ver- ification Office) tests of DSC Communications’ first generation Measurements of the ‘old’ cable type have shown that there will of WDM equipment. The original FVO test programme was be no noticeable increase in fibre attenuation at -20°C or above. supplemented, by including additional measurements required However, below -20°C the attenuation at 1550 nm will increase for the task to serve as a formal Telenor type approval test. The considerably, depending on individual fibre properties. The per- latter would largely comprise parameters which are listed in the formance during the winter months in the Hønefoss–Gol area, new ITU-T Recommendation G.692, however, without any spe- amongst the coldest in Norway, will therefore to some extent cific values attached to them yet. rely on effective temperature isolation, in particular on snow depth. The test programme was carried out partly at the premises of DSC in Denmark, partly as a field trial in Norway. The cable route between Oslo and Bergen, approximately 520 km, was chosen for the field trial. Six optical in-line amplifiers are being Table 1 Section lengths of the cable route Oslo-Bergen used, and no regeneration between the two terminals is required.

The FVO tests commenced on 30 June 1997 in Denmark, and Section Section name Length Length was successfully completed in Norway on 26 August, including no. (km) (dB) a 31 days on site stability test. 1 Oslo(Løren)–Gran 75.6 23.9/21.8 The Oslo–Bergen system is now in operation, carrying traffic. 2 Gran–Sokna 79.0 21.7/21.7 Although tested with four channels, provisionally only one is 3 Sokna–Nesbyen 76.1 20.6/20.3 now equipped. 4 Nesbyen–Geilo 71.4 20.6/18.1 The optical ADM is scheduled for testing first week of Decem- 5 Geilo–Myrdal 81.7 21.3/21.6 ber 1997. Following the approval tests, it will be installed on a second WDM route, Gjøvik – Trondheim. 6 Myrdal–Bulken 60.0 13.8/14.0 7 Bulken–Bergen 76.5 18.7/19.0 This paper briefly describes the FVO trial routes and the WDM equipment, subsequently presents the main results from the Total 520.3 expanded FVO technical report.

WDM TX1 WDM RX1

WDM TX2 WDM RX2 ILA OADM WDM TX3 WDM RX3 MUX DMUX WDM TX4 WDM RX4

Figure 1 Example of a simple point-to-point WDM system

Telektronikk 3.1997 115 On the other hand, FVO tests proved that the system margins The DSC WDM product concept, with respect to attenuation are considerable. In any case, the ‘old’ cable is considered to be replaced by double-window as delivered to Telenor SSMF. The DSC FOCUS WDM is a new product family, to be integrated with the DSC FOCUS AC4 SDH product line. The following system configurations will be supported: WDM point- to-point, WDM add/drop and WDM branching systems. Table 2 Remote measurements of optical parameters The first releases feature up to four WDM channels. A network Sokna example is illustrated in Figure 1 (only one direction is shown). Direction East–West West–East • The optical transmitters (TXs) use wavelength selected lasers Supervisory system NM2100 NM300 NM2100 NM300 according to the ITU-T G.692 wavelength grid (400 GHz spacing, 1549 – 1559 nm). Total input power (dBm) -12 -12.0 -9.5 -9.5 • The optical multiplexer (MUX) couples and amplifies the four channels. Total output power (dBm) +10 +10.0 +9.5 +9.5 • The in-line amplifier (ILA) operates with a constant total out- Laser bias current (mA) 162 162 142 142 put power. Two alternative optical interfaces are available, Laser temperature (°C) 25 25 24 24 22 dB and 33 dB nominal gain, corresponding to section lengths of approximately 80 km and 120 km, respectively. Pump power (mW) 43 43 40 40 The 22 dB ILA is a two-way optical repeater, using one separate amplifier for each direction of transmission. • The optical add/drop multiplexer (OADM) is bidirectional, dropping and adding channels in both directions, using fixed Table 3 Measurements of central wavelengths optical filters.

Nominal DSC ITU-T range Oslo–Bergen Bergen–Oslo • The optical demultiplexer (DMUX) – combined with the wave- tolerance (nm) (nm) (nm) (nm) MUX in one single module – just amplifies and splits the length Temp./Life incoming WDM signal. (nm) • The optical receivers (RXs) incorporate tuneable optical filters for dynamic optical demultiplexing. 1549.32 ± 0.2 1548.67 – 1549.96 1549.7 1549.8 The WDM product family also includes other modules (optical 1552.52 ± 0.2 1551.88 – 1553.17 1553.0 1553.0 transponders, demultiplexers with fixed filters, optical branch- 1555.75 ± 0.2 1555.10 – 1556.39 1556.2 1556.2 ing), but these were not tested. 1558.98 ± 0.2 1558.33 – 1559.63 1559.5 1559.4 For supervision and management of the WDM elements, the DSC NM2100 network management system can be used, as well as NM300-WDM as a local craft terminal.

Trial routes configurations

For the trial routes at DSC and on site in Norway, four 2.5 Gbit/s channels (double sets of optical transmitters/ receivers) were equipped. At both locations also six 22 dB ILAs were being used.

The link at DSC was equivalent to the Oslo–Bergen route, except that in one direction of transmission optical attenuators were inserted instead of fibre drums between amplifiers.

The DSC system is capable of coping with a fibre dispersion of up to 10,000 ps/nm, corresponding to a total system length of approximately 540 km.

Test of remote supervisory system

A set of five optical parameters per direction, from each ILA and each MUX/DMUX, can be measured. Figure 2 Spectral characteristics of optical transmitter No. 3 in Oslo

116 Telektronikk 3.1997 An example is shown in Table 2. The ILA of ‘Sokna’ was at first checked at DSC, using both NM2100 (the network management system, connected to MUX/DMUX in ‘Oslo’ via the Terminal Server interface box) and NM300 (local craft terminal, connected to the ILA of ‘Gran’). Obviously, the results are corresponding.

NM2100 is also capable of forcing-off/switching-on individual lasers, LOS will be reported to both NM2100 and NM300.

Central wavelengths 3a Test instrument used: W&G OMS-100 Optical Mea- suring System (calibrated). All measurements were performed in Oslo. Results are given in Table 3. All values measured are relatively high, above nominal figures. They are outside DSC tolerance figures, however, within the ITU-T recommended range. The corresponding measurements previously carried out at DSC, using the optical spectrum analyzer, were close to nominal wavelengths. Obviously, test results may differ more than expected when using different test instruments, although both calibrated.

Spectral characteristics of optical transmitters

Test instrument used: Ando AQ-6315A Optical Spec- trum Analyzer. 3b As laser modulation bandwidth is in the order of 0.02 nm, the -20 dB width could not be measured using Figure 3a/3b Spectral characteristics at MUX output in Oslo the optical spectrum analyzer.

For the scanning a resolution of 0.2 nm and Mode ‘High 1’ was used. (This mode suppresses noise and performs a more correct measurement of the optical signal. Measurement duration is also convenient.)

All optical transmitters of the ‘Oslo’ terminal station were measured at DSC. The spectral characteristics of transmitter No. 3 is shown in Figure 2, as an example.

Test results ‘Oslo TX3’: Central wavelength 1555.440 nm Power level -2.67 dBm Highest side mode 1553.380 nm Power level -44.27 dBm Wavelength difference -2.060 nm Power difference -41.60 dB

Vacuum measurements were also performed, resulting in a central wavelength of 1555.88 nm, i.e. 0.44 nm higher. (The ITU-T Recommendation G.692 refers to vacuum, nominal value and range are shown in Table 3.) Figure 4 Optical spectrum at Bulken ILA output

Telektronikk 3.1997 117 Table 4 Individual channel power level (dBm) at ILA outputs and Bergen DMUX output

ILA name Gran Sokna Nesbyen Geilo Myrdal Bulken Bergen ILA no. 123456DMUX

1548.98 nm 2.69 2.22 1.34 1.52 0.73 0.82 -12.04 1552.24 nm 2.78 2.60 1.90 2.07 1.14 1.59 -10.81 1555.44 nm 3.29 3.55 3.56 4.06 3.30 4.29 -7.81 1558.76 nm 3.64 3.85 4.00 4.25 3.18 4.29 -7.82

Max ch. power diff. 0.95 dB 1.63 dB 2.66 dB 2.73 dB 2.57 dB 3.47 dB 4.23 dB

Side mode suppression ratio for the four transmitters in Oslo were found to be: Oslo TX1: 45.72 dB Oslo TX2: 47.94 dB Oslo TX3: 41.60 dB Oslo TX4: 46.10 dB

DSC requirement: SMSR = 30 dB ITU-T G.957 recommendation for STM-16: SMSR = 30 dB.

Central wavelengths are all lower than the values later to be measured in Oslo (using a different test instrument).

Optical spectrum along the route

The optical spectrum was recorded at MUX output, all ILA outputs, DMUX output and DMUX test out- Figure 5 Optical spectrum at Oslo MUX/DMUX Test Out (direction Oslo–Bergen) put. For practical reasons this test was carried out at DSC, and in Norway only repeated in Oslo MUX/DMUX. Oslo MUX output, as measured at DSC, is shown in Figure 3a (1545 – 1560 nm) and 3b (1250 – 1650 nm). In Figure 3b one will see, from left to right: the OSC (optical supervisory channel) at 1310 nm, the optical pump power at 1480 nm and finally, the four traffic channels. The per channel power at Oslo MUX output was measured to be (using the optical spectrum analyzer): 1548.980 nm: +6.78 dBm i.e. 4.76 mW 1552.240 nm: +6.32 dBm i.e. 4.29 mW 1555.440 nm: +6.12 dBm i.e. 4.09 mW 1558.760 nm: +6.38 dBm i.e. 4.35 mW In total +12.40 dBm (17.37 mW) Figure 6 Optical spectrum at Oslo MUX/DMUX Test In (direction Bergen–Oslo) i.e. in total 17.49 mW

118 Telektronikk 3.1997 Figure 7 Eye diagrams at optical transmitter outputs in Oslo

Maximum channel power difference: 0.66 dB. SNR > 35 dB. Results are shown in Figure 5 (Oslo MUX Test Out – i.e. direction Oslo–Bergen) and Figure 6 (Oslo DMUX Test In – i.e. Moving from Oslo to the sixth ILA, Bulken, the optical direction Bergen–Oslo). From these results one may conclude: spectrum has changed some, as one might expect. Please refer to Figure 4. SNR is clearly better than 35 dB for all outgoing channels, in the area 25.5 – 28 dB for all incoming channels. Channel power levels, at the output of each ILA, are shown in Table 4. Maximum channel power difference is 5.2 dB, i.e. slightly more than what was measured at DSC, direction Oslo-Bergen. The spectrum diagram at Bergen DMUX output indicates an SNR > 24 dB. This means that from MUX output in Oslo to DMUX output in Bergen SNR has decreased 10 – 13 dB. Eye diagram and extinction ratio Eye diagrams of all optical transmitter outputs (measured in As can be seen from Table 4, maximum channel power differ- Oslo) are shown in Figure 7. ence increases along the route (albeit irregularly), reaching a value of 4.23 dB in Bergen. This power difference is no prob- All values of extinction ratio (ER) range from 9.1 to 14.0 dB. lem to handle (10 dB can be accepted). According to DSC product specification, the minimum value for ER should be 8.2 dB. (This figure is in agreement with the The optical spectrum measurements performed on site in Oslo, ITU-T requirement for 1550 nm V-type STM-16 interfaces for made use of an Advantest Q8381 Optical Spectrum Analyzer.

Telektronikk 3.1997 119 Figure 8 Eye diagrams at optical receiver inputs in Oslo

single channel systems. There is not yet any requirement in e.g. attenuation in three sections was set to 10 dB simul- G.692.) taneously. Test time varied from 3 minutes to 15 minutes, no bit errors were reported. Eye diagrams of all optical receiver inputs (measured in Oslo) are shown in Figure 8. A BER of 10-12 corresponds to approximately 0.56 errors per hour, or 13.44 errors per 24 hours referring to STM-1. The The four eye diagrams are noticeably different, as also was indi- measurements above are therefore indicative only. cated on transmit side. Obviously, the Oslo–Bergen route seems to be approaching the maximum systems length. Amplifier section attenuation was also increased. Section 7 could accept 17 dB additional attenuation, no errors in a 7 minutes test period. Section 1 could accept 16 dB additional System margins attenuation, no errors in a 3 minutes test period. Section 5 could accept 9 dB additional attenuation, no errors in a 3 minutes test Attenuation margins were tested at DSC, using optical attenu- period. ators instead of cable on all the seven amplifier sections. Various combinations of reduced section attenuation were tried,

120 Telektronikk 3.1997 Finally, attenuation was increased simultaneously in 4 sections, numbers 1, 2, 5 and 7. The total increase corresponded to 35 dB beyond the nominal maximum (22 dB x 7 = 154 dB). No errors were observed during the test period. The margins seem to be pretty safe with respect to attenuation. Lillehammer Myrdal To investigate dispersion penalty, a simpli- Voss Gol Ål Gjøvik Hamar fied test was carried out in Norway. It was Geilo concluded that penalty caused by cable Jevnaker dispersion was negligible. Bergen Brandbu Kongsvinger 31 days stability test Hønefoss Oslo Lillestrøm The Oslo–Bergen system was looped in Drammen Bergen on STM-1 level and supervised for one month of operation. No errors occurred during this period. This result may seem Horten Moss satisfactory. Fredrikstad Porsgrunn Stavanger Other tests Larvik

Automatic laser shutdown and restart functions (ALS and ALR) were verified. Output power Arendal from optical transmitter alternated between ‘on’ (-2.5 dBm) for 2 seconds and ‘off’ for 70 seconds after Kristiansand disconnection. When applying manual restart, 2 seconds pulses and 90 seconds pulses (for measurement purposes) Figure 9 Cable route from Oslo to Bergen respectively, were checked.

Basic alarm test was performed.

Some module tests were left out, as a test report prepared by DSC contained the necessary information, and there did not seem to be a reasonable need for repeating them.

Conclusion

An overall evaluation of the test results easily concludes that DSC FOCUS AC4 Release 2.5 WDM equipment can be approved for application in Telenor’s network. Performance has been proven to be satisfactory, partly beyond expectations, and key system margins seem to be more than adequate.

Acknowledgements

The author wishes to acknowledge all who participated in the test programme, • from DSC Communications A/S: Aiguo Li, Carsten G. Jørgensen, Martin Seerup Nielsen, Ole M. Olesen • colleagues from Telenor AS: Steinar Bjørnstad, Ottar Flydal, Steinar M. Svendsen is Chief Engineer at Irene Kaasbøll and Jon Tjøgersen. Telenor Nett AS, Infrastructure Division. Since 1983 his major area of interest has been optical line transmission systems for the trans- Also appreciated is the kind permission granted by DSC Com- port network. He is currently working within the munications to publish product and measurement material con- SDH project of Telenor. tained in this paper. e-mail: [email protected]

Telektronikk 3.1997 121

122 Telektronikk 3.1997

Status International research and standardization activities in telecommunication

Editor: Per Hjalmar Lehne

Telektronikk 3.1997 123 124 Telektronikk 3.1997 Introduction PER HJALMAR LEHNE

In this issue of the Status section we take a closer look into main objective is to define techniques for the integration of the some projects from the ACTS (Advanced Communications, satellite component into UMTS. Special focus is on the inter- Technologies and Services) programme. ACTS has a total working with the terrestrial component. As in all ACTS pro- budget of 670 MECU, or 5 % of the total budget for the Euro- jects, trials play an important role also here. pean Union Fourth Framework Programme. The programme is divided into six technical domains and one supporting, or horizontal, domain: Table 1 ACTS projects with Norwegian participation • Interactive Digital Multimedia Services • Photonic Technologies Interactive Digital Multimedia Services • High-Speed Networking • Mobility and Personal Communication Networks CRABS Cellular Radio Access for Broadband Services • Intelligence in Networks and Services MAESTRO Maintenance System based on Telepresence for Remote • Quality, Security and Safety of Communication Systems and Operators Services • Horizontal Actions. Mobility and Personal Communication Networks MoMuSys Mobile Multimedia Systems The ACTS programme is probably the most important driving SAMBA System for Advanced Mobile Broadband Applications force in Europe into the future of telecommunications technologies. SINUS Satellite Integration into Networks for UMTS STORMS Software Tools for the Optimization of Resources in Mobile Telenor is currently participating in 17 projects in ACTS, Systems additionally there are Norwegian participation in three more projects. The projects are listed in the table. Four of the projects High-Speed Networking are presented here. ASICCOM ATM Switch for Integrated Communication Computation and Monitoring First, Mr. Agne Nordbotten describes the CRABS project (Cellu- lar Radio Access for Broadband Services). This project focuses CA$HMAN Charging and Accounting Schemes in Multi Service ATM Networks on interactive services operated at 40 GHz. Trials will be performed in five European countries. It is expected to form a EXPERT Platform for Engineering Research and Trials foundation for future system recommendations. In this project, JAMES Provision of European Networking Facilities (PEN) Telenor has the leadership as Prime Contractor, with Mr. Nord- botten as project leader. In addition to describing the project NICE National Hosts Interconnection Experiment objectives, he also shares some of his experiences of having the REFORM Resource Failure and Restoration Management in ATM- leadership and the benefits for Telenor. based IBCN SMASH Storage for Multimedia Applications Systems in the Home Mr. Torodd Olsen has, together with the other partners, contributed with a description of the OPEN project (Optical Pan- TRUMPET TMN's Regulations and Multiple Providers Environment European Network). This project focuses on the possibilities of Photonic Technologies using an overlay network based on optical fibres and transparent optical cross-connects (OXC). Field trials will be performed BLISS Broadband Lightwave Sources and Systems between Norway and Denmark and between France and Bel- MEPHISTO Management of Photonic Systems and Networks gium. The main objective is to assess the feasibility of the concept. Telenor’s role in the project is also presented. OPEN Optical Pan-European Network Quality, Security and Safety of Communication Systems and Services Multimedia, i.e. joint use of text, audio and video to provide information, or as a communications means, is a growing field. RETINA An Industrial-Quality TINA-Compliant Realtime DPE This is made possible by the merging of advanced computing and information processing and of broadband telecommunica- Horizontal Actions tions technologies. In the near future, multimedia services will INFOWIN Multimedia Information for National Hosts also be available when mobile. Mr. Robert Danielsen presents OPTIMUM Optimized Network Architectures for Multimedia Services the MoMuSys project (Mobile Multimedia Systems) which aims at the development of mobile multimedia technology. The project is very strongly linked up to standardisation work in ITU1 and ISO2. He describes the results up to now and also has some Per Hjalmar Lehne is Research Scientist at thoughts about how the work can be continued. Telenor Research & Development, Kjeller. He is working in the field of personal communications, with a Finally, Ms. Vendela Paxal presents the SINUS project (Satellite special interest in antennas and radio wave propaga- Integration into Networks for UMTS3 Services). This project’s tion for land mobile communications. email address: [email protected]

1 International Telecommunications Union, a UN body. 2 International Standardisation Organization. 3 Universal Mobile Telecommunications System.

Telektronikk 3.1997 125 The ACTS Project AC215 CRABS; Cellular Radio Access for Broadband Services AGNE NORDBOTTEN

This article gives a presentation of the ACTS project AC215 and equipment studies including technical trials. The CRABS which focuses on the development and demonstra- project is highly trials oriented and trials have now started tion of a broadband radio system for interactive services in five European countries. The final results of the project operated at 40 GHz. The project covers service and user are expected to form the foundation for recommendations of aspects, architecture and systems studies including extensive a future system. simulations as well as studies of propagation issues to determine propagation type QoS criteria and coverage models 1 Introduction

The main objective of this project is to develop and demonstrate a broadband radio access system with interactivity operated in Table 1 Partners of the CRABS project the frequency range 40.5 – 42.5 GHz which is the frequency range allocated in most European countries. The results from Code Partner name Partner roles these trials in combination with systems and propagation studies as well as evaluation of technology, equipment and user and P01 Telenor R&D, Norway Prime contractor, WG leadership, service requirements, will form the basis for the final systems Trial site recommendations. The project is thus focusing on extensive P02 Philips Broadband Networks, Ind. partner, WG leader technical, user and service trials performed at different locations UK in five European countries, systems studies and simulations, establishment of propagation and coverage models to be used AP02A Philips Research Laboratories, Associate partner P02, Res.org. for network planning as well as evaluation of results and recom- Redhill, UK mendations. AP02B Philips SpA, Monza, It Associate partner P02, Res.org. The project started July 1996 and has a duration of 30 months P03 Rutherford Appleton Labo- WG leadership, Propagation trials with a total effort of altogether 1017 man months divided ratories, UK among 9 partners and 3 associate partners as listed in Table 1. P04 Joanneum Research, Research organisation associated This paper presents the philosophy and objectives of the project, Austria with the University of Graz the main tasks addressed in the different workgroups, the ob- P05 TESTCOM, Czech Rep. Research Organisation, Trial site jectives of the trials and some preliminary conclusions from the first trials. P06 RAI, It Broadcaster, Trial site AP06A Politico di Milano, It University, associated with P06 2 Background and basic systems aspects P07 EUROBELL, UK Cable operator, Trial site Originally, the frequency range 40.5 – 42.5 GHz was planned to P08 DEMOKRITOS, Greece Research organisation, Trial site be used for broadcasting or redistribution of broadcasting, a P09 THOMSON CSF, Fr Industrial organisation wireless cable system. As the cost of cable provision was estimated to be between 20 – 40 ECU per metre, depending on local conditions, a significant number of rural and suburban homes with long feeder connections could be denied broadband access to Broadcast Channel interactive services. However, the event Tx Rx of digital broadcasting by satellite opened for new broadcasting related ser- Radio- Tx vices including interactivity and integra- Return tion of data and telecommunication ser- QPSK Channel vices. The growth of Internet traffic Modulator clearly demonstrated the possibilities of Signal new interactive services. It became Set Top Box from widely recognised that broadband inter- Satellite active connections to the homes and to Multi- smaller enterprises and organisations plexer would be required and that a radio based solution could be very attractive and TV PC flexible when sufficient bandwidth was available.

Server The project discussions started at the very beginning of digital broadcasting by satellite and during the first Internet Control wave. It seemed obvious that services Terrestrial Return Channel based on a combination of broadcasting, interactivity, data exchange and new Figure 1 Sketch of basic system

126 Telektronikk 3.1997 very flexible telecom services, could have an almost enormous 4 The CRABS trials growth potential into the beginning of the next century. It was felt that a broadband system based on the new digital broadcast Since ACTS is a program focusing on trials and demonstrations, technology, including data, for the downlink and a flexible use this is also a central part of CRABS with trials being performed of communication technology for the uplink, would offer the in five different countries on the premises of the different part- possibility of creating a cellular access system for real broad- ners. While most ACTS projects have more or less centralised band services in the local domain. A systems solution along trials which very often are located at National Hosts, the these lines is outlined in Figure 1. CRABS trials are distributed throughout Europe and there is a local motivation behind each trial site depending on geographi- The solution is based on the application of the MPEG2/DVB cal location, cultural, economical and technical background, technology used for satellite broadcasting including data of any type of operator involvement and level of expectation. Within kind. Thus, also QPSK modulation is being used in order to be the project it is felt that this is the best way to create an interest able to use the digital Set Top Boxes developed for satellite for and to test new services. reception. At the receiver site the 40 GHz signal is down- converted from 40 GHz to the intermediate frequency range Regarding the technological background represented by factors 950 – 2050 MHz of the satellite receiver. Reception is presently like the density of PCs in the homes and the volume and growth based on this type of receivers in combination with PCs. It is of interactive services, there is a strong gradient from the north very important that the user interfaces required for services to the south of Europe, with the Nordic countries in the lead. using different networks are identical to the extent possible. At the transmitter side information signals are multiplexed on to Different types of services have been chosen for the trials, MPEG2 transport streams and transmitted in the frequency rebroadcasting of digital satellite programs, video on demand, range 40.5 – 42.5 GHz. The signals may contain retransmitted Internet, information services, conferencing and educational TV signals, data obtained from servers through interactivity like services to mention some. Internet services, and on-line services like conferencing. Inter- activity may be obtained through available terrestrial technol- Interactivity is definitely an important issue and the project has, ogy like ISDN, DECT, or radio relay systems. At the end of the at least from the start, adopted the interactive reference model project an in-band return channel at 40 GHz will be included. developed by the ACTS projects INTERACT and DIGISAT. This model is shown schematically in Figure 2.

3 The partners and their roles The downlink channel is a broadcast channel or point to multi- point channel containing the information addressed to groups The basic project ideas originated in a small group of partici- and individuals. The coverage range is typically 3 – 5 km with pants with three of the final partners as dominant contributors. an opening angle of 50 – 60°. The uplink, interactive or return When the structure was more or less worked out the need for channel is a typical communication channel where different additional partners was discussed from both a technical and technologies may be used depending on availability and the political point of view. The decision was to make a winning needs of the user. proposal and it was known that the competition within this field would be very strong and that several proposals were expected. The CRABS trials are divided into two phases running for app- The proposal became a winner with the partners listed in Table 1. roximately one year each. In phase 1 available terrestrial technology like ISDN, DECT and radio relay technologies are being This clearly illustrates the broad and strong combination of used depending on trial site, while phase 2 will also include in- operators in the project with Telenor representing the telecom, cable and satellite sides, EUROBELL is a cable operator interested in investigating the possibilities of employing broadband radio technology to extend their existing services and to add new flexible interactive services. The broadcaster RAI is looking for new activities in the local domain and is in this pro- Service User ject testing educational aspects, while TESTCOM has the much Providers Terminal broader view of introducing digital services as soon as possible in the Czech republic. For DEMOKRITOS the introduction of Broadcast interactive services is the main objective. Channel Delivery medium Broadcast Broadcast Satellite, Terrestrial Module Industry is represented by Philips and THOMSON CSF, LMDS, Cable (TV) Rutherford Appleton Laboratories is recognised as the leading institute on propagation issues in Europe. Joanneum Research is known for contributions to satellite communication through ESA, leadership of COST projects and recognised competence in the field of transmission protocols. Politecnico di Torino, PSTN/ISDN Interactive Service associated with RAI, are specialists on microwave and optical Interactive GSM, DECT Satellite Module technology. (PC) Interactive Radio, LMDS Thus, the project is well composed with regard to the work to be Channel done. Figure 2 Interactive reference model

Telektronikk 3.1997 127 band radio return at 40 GHz. Telenor R&D is using ISDN 5 The role of prime contractor connections for the first phase concentrating on testing the broadcast channel and interactivity in itself with a high capacity In this project Telenor has also had the role of prime contractor. delivery channel. It was not the intention during the planning period, but suddenly the situation changed, and the challenge was accepted. At the trial site in Kent, UK, the cable operator performing the trials is hoping to extend the business potential and a radio path This role definitely takes a lot of effort, and it is very different offering high capacity is wanted in order to offer broadband ser- from a local project where all the participants are well known vices to smaller enterprises and schools. The Telenor trial at and the project leader can organise the project with that as a Kjeller is directed towards five selected users/user groups. The background. In an international project it takes some time to interactive channel during phase 1 is ISDN based, and an in- obtain such background knowledge, and at that time the differ- band RF channel will be tested in phase 2 starting next year. In ent activities in the project are being addressed, and not neces- Italy RAI is performing trials/demonstrations at different loca- sarily in an optimised way. The creation of motivation in such a tions in Venice, Turin and Rome. The Rome trials are mainly project can definitely be a difficult task. In the CRABS project educational. The Greek trials are directed towards tourist in- we have been lucky as we have so far avoided destructive and formation purposes. Through connections with databases local lengthy discussions on minor topics. information is obtained or connections can be established into Internet where local newspapers etc. from home countries may The prime contractor also has the responsibility of representing be obtained. the project towards Brussels, which is an advantage at least when the results from the project are reasonable good. It gives The trials are expected to give: the possibility for active participation in domain meetings and workshops where the flow of information from other projects is • Information on equipment needs considerable, a forum for information exchange and concen- • Variations of QoS with meteorological/topological conditions trated watcher activities. Presentations of the project also represent possibilities for international contact and recognition. • A possibility for testing of both architectural models and coverage models • A background for contributions to standardisation of systems 6 Conclusion and services The participation in the CRABS project has given us valuable • A testbed for ideas on how broadband services will develop experience in leadership of relatively large international pro- in Europe jects. Access to the international research community in this way is very challenging from a professional point of view. The • The views from different operator groups and different loca- interest for work on broadband radio access has been growing tions around Europe regarding new broadband services. during the months the project has been running, a fact which is also reflected in the increasing interest for the project. It is So far the results look promising. Trials are now started in all expected that the last 15 months of the project will show good five trial countries. The main obstacle up to now has been lack and interesting results from trials, systems and propagation of stability in parts of the transmitter equipment. This has to be studies and that these results will contribute heavily to the de- improved. The studies for further development of the system as velopment of radio based broadband interactive services. The well as the propagation studies are giving good results. time from trials to real services seems to be short in this field. Except for some equipment problems at the start, the project got up to a proper speed in a short time and is now moving according to the intentions.

Agne Nordbotten is Research Manager Satellite and Radio Systems, Telenor Research & Devel- opment, Kjeller, and Project Coordinator for AC215 CRABS. email address: [email protected]

128 Telektronikk 3.1997 AC066 OPEN – Optical Pan-European Network TORODD OLSEN

1 Main objectives into layers in a way that the number of hops required in the interconnection of the two The ACTS project OPEN aims at the study of the feasibility of nodes is minimised. In addition, a fourth an Optical Pan-European overlay Network (OPEN), inter- approach, termed a modular/hybrid topol- connecting major European cities by means of a mesh of high- ogy, is also being studied. Furthermore, capacity optical fibre links cross-connected through transparent the project is developing an end-to-end photonic nodes. The network is built on an existing fibre infras- transmission model based on node and OPEN tructure, and offers an upgradable capacity. As can be seen in link modelling activities. Figure 1, the scope of OPEN addresses the transport in the optical transmission medium layer and the optical path sub-layer. The potential capacity of each fibre link will be upgradable at 3 Enabling technologies least up to 40 Gbit/s, each channel supporting STM-16 SDH or In the OPEN network concept, all optical WDM higher data-rate transport service. The proposed approach relies routing is implemented by means of multiwavelength on the extensive use of wavelength division multiplexing optical cross-connects (MOXCs), located at major European (WDM) for both transmission and routing purposes, but the cities and interconnected via WDM links. All-optical WDM interfacing with high-speed trunks carrying optical time- technologies are used in this concept both for transmission be- division multiplexed (OTDM) channels are considered as well. tween the network nodes and routing through MOXCs, exploit- The goal of the project is to assess by modelling, laboratory ing exclusively optical regeneration and reshaping of incoming experiments, and field implementation of a photonic network data. The WDM components for transmission are flat-gain opti- prototype, the feasibility of the OPEN concept [3,4]. cal amplifiers and wide-band dispersion compensating devices. The main requirement on these components is a wavelength- 2 Network concept and architectures insensitive operation in the entire spectral band of the multiplexed channels. On the other hand, the principal functional The advantages brought by the use of WDM for transmission blocks of the MOXC are a multiwavelength space switch, a and routing within a network like OPEN are numerous. First, tuneable optical filter, an all-optical wavelength converter and a this technique allows for an increase of capacity of each link, flat-gain optical amplifier. Figure 3 shows the architecture of without resorting to the use of ultra fast optical technologies. In the OPEN MOXC, which is based on the broadcast-and-select addition, this increase of capacity is accompanied by a higher principle. The incoming multiplexed signal from each port is resistance to both group velocity and polarisation mode dis- split as many times as the number of output wavelength chan- persion of the line fibre, which limit the transmission distance nels. For each output channel, the correct input port containing through the individual channel rate. Therefore, WDM provides the relevant input channel is selected at the switching stage. At non-regenerated span x throughput products that are substan- the tuneable filter stage this channel is selected from the input tially higher than other techniques. Second, WDM allows for an multiplexed signal and, if necessary, its wavelength is translated enhancement of flexibility of routing techniques, by the use of before multiplexing into the outgoing fibre. the wavelength dimension. In this context, all-optical WDM routing techniques in the nodes offer the possibility of imple- The all-optical wavelength converter is a non-linear device that menting simple network configuration, protection, restoration provides some signal reshaping such as extinction ratio en- and management schemes. Finally, and perhaps most im- hancement, limitation of noise accumulation and spectral portantly, all-optical WDM routing techniques provide such a network with a high degree of transparency: any type of service, with any bitrate, and adopting any digital frame format, can be handled and transported from one end of the network to the End System other with the same optical equipment, avoiding Circuit Layer the need of electrical regeneration in the nodes. Switch

In terms of network analysis, the project has established the full description of the network, Digital Path Sub-Layer including the overall dimensioning, the evolutionary path from the existing infrastructures, the DXC DXC operation and management aspects and the pro- Bearer Network tection schemes, so as to propose a sound vision of the various steps leading to the implementa- Optical Path Sub-Layer OXC OXC

tion of the OPEN network in a medium-term and Layer Transport in a longer-term perspective. Four topologies have been studied for the network, as depicted in Figure 2. The first one is the flat topology, which optimises the position of the OXCs with respect Scope of the to traffic density and capacity. The second one is OPEN Project the partitioned topology, which seeks to partition Optical Transmission Medium Layer the network into WDM islands, which may then be interconnected with OTDM high-speed trunks. The hierarchical topology subdivides the network Figure 1 OPEN network concept

Telektronikk 3.1997 129 Oslo Helsinki Secondary Nodes Oslo Helsinki Stockholm Primary Nodes Stockholm

Dublin Dublin Copenhagen Copenhagen Amsterdam Amsterdam London Berlin London Berlin Brussels Brussels Paris Lux Prague Paris Lux Prague Vienna Vienna Zürich Zürich

Zagreb Zagreb

Lisbon Madrid Rome Lisbon Madrid Rome

Athens Athens a) FLAT Topology b) PARTITIONED Topology

Helsinki Helsinki Oslo Oslo Stockholm Stockholm

Dublin Dublin Copenhagen Copenhagen Amsterdam Amsterdam London Berlin London Berlin Brussels Brussels Paris Lux Prague Paris Lux Prague Vienna Vienna Zürich Zürich

Zagreb Zagreb

Lisbon Madrid Rome Lisbon Madrid Rome

Athens Athens c) HIERARCHICAL Topology d) MODULAR Topology

Figure 2 The four topologies being studied in OPEN

cleaning, allowing to overcome several physical limitations based EDFAs were used. The fluoride-based EDFAs provide a (like noise accumulation, cross-talk in switches and filters, filter relatively flat gain over the wavelength range 1530–1560 nm, bandwidth narrowing, chromatic dispersion and non-linear allowing large channel spacings, compatible with field require- effects in the transmission spans) that would otherwise drasti- ments. Here, 16 channels separated by 200 GHz and spanning cally limit the achievable network expansion. The OPEN cross- 24 nm (between 1536.61 and 1560.61 nm) were used. The connect is implemented using some advanced optical com- receiver sensitivities at 10-10 BER for the 16 channels show a ponents, among which are a semiconductor-based optical switch total excursion of 3.1 dB. optimised for multiwavelength operation and an all-optical interferometric wavelength converter. Node cascadability experiments are also carried out aiming to assess the feasibility of a meshed Pan-European network. The The laboratory experiments carried out by OPEN are using OPEN project carried out laboratory experiments with the cross- recently developed advanced optical components. The project connect architecture described in the previous section, with the has demonstrated successfully the transmission of 16 x 10 Gbit/s use of the advanced technological components provided by the signals over 531 km of standard fibre, using dispersion manage- project. In particular, a nearly penalty-free cascade of three sub- ment [1]. These experiments featured Dispersion-Compensating equipped optical cross-connects with a total transmission dis- Fibre for dispersion management and only 7 in-line fluoride- tance of 1000 km on standard G.652 fibre was demonstrated,

130 Telektronikk 3.1997 1 1

m λ λ λ1... λn 1 1... n thanks to the reshaping properties of the all-optical mn interferometric wavelength converter [2]. The used n cross-connects were representative for 4 x 4 size cross-connects, with 8 wavelengths per port and 2.5 Gbit/s per wavelength. λ λ λ1... λn 1... n 4 Field trials – the role of Telenor

OPEN will carry out two demonstrations in two different cross-border field trials, the first one between Norway and Denmark, and the second one between France and Belgium. The schematic layout of the m trials is given in Figure 4. STM-16 signals will be transmitted and routed, using 5 channels (4 working channels and one protection channel), 400 GHz apart. Under the constraint of a limited infrastructure length, the choice was made to implement a node physical connectivity of 3 or 4 with unidirectional Local A/D input/output ports links rather than using fully bi-directional links. The OPEN project will demonstrate the powerfulness of its approach in using the same cross-connect, ampli- space selector fiers and chromatic dispersion compensation units in the two different field trial environments. In both wavelength selector field trials long distance WDM transmission and wavelength converter routing experiments, as well as network restoration demonstration, will be performed. Configuration, multiplexer performance, alarms and protection management will be demonstrated. Some important demonstra- Figure 3 OPEN MOXC architecture tions are planned: in the N-DK field trial, critical issues are four wave mixing (FWM) effects and very long repeater spacings in the two hybrid G.652/ G.653 links bridging Arendal to Thisted either France - Belgium Trial Norway - Denmark Trial through Kristiansand or through Hjørring; in the F-B field trial, the span of WDM transmission and rout- Brussels Oslo ing experiments can reach more than 1500 km with a (B) (N) cascade of four wavelength translating cross-connects in addition to other types of nodes (wavelength 228 km 115 km 350 km routing cross-connect, crossbar-switch). Fast restora- (G652) (G652) (G652) tion is another issue for this field trial.

The role of Telenor is in the network studies and pri- Lille Arendal marily in the planning and facilitation of the Nor- Amiens (F) (N) way-Denmark field trial. Prior to the field trial, a (F) comprehensive measurement programme has been 113 km carried out to determine the fibre parameters in terms (G652) 200 km 140 km (G652/G653) (G652) of attenuation, chromatic and polarisation mode dispersion. In particular, much attention has been paid 149 km to the study of the effect of FWM in the dispersion (G652) shifted fibres in submarine cables. Determination of the zero dispersion wavelength has been an important input parameter to the analysis of this subject. Hjørring (DK) The measurement data from the fibre characterisa- Paris Thisted tion have been fed into transmission models pro- (F) (DK) vided by several partners to simulate the field trial network performance and optimise network design, Figure 4 Schematic representation of the OPEN field trials such as amplifier spacing and output power levels. Significant contributions have been made by SIN- TEF, the sub-contractor of Telenor R&D, in the study of the Pan-European network architecture, transmission limitations node in Oslo, a central node with the optical cross-connect and and analysis of field trial network performance. add-drop facilities located in Arendal, a transit node in Hjørring with a channel drop option, and finally, the terminal node in Figure 5 shows the network elements and fibre infrastructure of Thisted. The OXC and the line amplifiers featuring state-of-the- the N-DK trial network. The network consists of a transmit art technology are supplied by Alcatel Alsthom Recherche,

Telektronikk 3.1997 131 ARENDAL HJØRRING

OSLO 140STM-16 99 RX STM-16 TX-1 OPEN OXC STM-16 M TX-2 U 55 47 66 66 69 53 AUX TX-3 X TEST KRISTIANSAND TX-4 77 R T 140 86 X X optical amplifier STM-16 STM-16 G652 RX STM-16 G653 THISTED Figure 5 Network elements and fibre infrastructure in the N-DK field trial

while the terminal equipment is provided by the telecom operators. An important feature of the trial network is the interface References to the existing national SDH-network, which is implemented by 1 Artigaud, S et al. Transmission of 16x10 Gbit/s channels means of STM-16 interface transponders developed by Telenor spanning 24 nm over 531 km of conventional single-mode R&D. All planning activities and operations related to the fibre using 7 in-line fluoride-based EDFAs. Conference on N-DK field trial are carried out in close co-operation with Optical Fibre Communications (OFC’96), San Jose, CA, TeleDanmark and SINTEF, and strongly supported by Alcatel February 26 - March 1, 1996, post-deadline paper PD27. Alsthom Recherche and Alcatel Telecom Norway. The N-DK field trial installations, evaluation programme and demonstra- 2 Jourdan, A et al. Experimental demonstration of penalty- tions are going to take place in the period October-December free optical crossconnect cascade with multiwavelength 1997. The OXC and the line amplifiers, supplied by Alcatel, are transmission over 1000 km standard fibre. Conference on subsequently shipped to the sites of the F-B field trial for com- Optical Fiber Communications (OFC ’97), Dallas, Feb. 97, plementary tests. paper TuR3. Both field trials are aiming at the validation of the future all- 3 Chbat, M W et al. WDM for transmission and routing in optical Pan-European network and proving the new technology wide-scale networks : the Optical Pan-European Network in real operating environments. approach (OPEN). NOC’96, vol. 1, Broadband Superhigh- way, 182–186, Heidelberg, June 1996. Acknowledgements 4 Chbat, M W et al. Pan-European Optical Networking using The author wishes to acknowledge the key managers of the Wavelength Division Multiplexing. IEEE Communications OPEN Consortium, M. Chbat, A. Leclert, E. Jones, B. Magazine, 35, (4), 1997. Mikkelsen, N. Flaarønning, T. van den Broek, M. Pouleo, H. Melchior, P. Demeester, P. Gødsvang, A. Jourdan, and D.R. Hjelme for contributing to the source material on which this List of participants in the OPEN article is based. Consortium

F Alcatel Alsthom Recherche N Alcatel Telecom Norway B Belgacom SA I Centro Studi e Laboratori Telecomunicazioni – CSELT CH Eidgenössische Technische Hochschule Zürich – ETH-Z F France Telecom – CNET Torodd Olsen is Research Scientist at Telenor Research & Development. He has been employed by B Interuniversity Micro-Electronics Centre IMEC Norwegian Telecom since 1976, apart from a period with Tandberg AS from 1980 to 1986, when he started N SINTEF at Telenor R&D working with optical high speed and DK Technical University of Denmark coherent communications systems. Since 1995 he has been Project Manager in optical communications, and DK TeleDanmark is currently manager of the Optical Networking group. N Telenor e-mail: [email protected] UK University of Essex.

132 Telektronikk 3.1997 MoMuSys ROBERT DANIELSEN

1 Introduction tele-presence to mobile multimedia communications, programme production and generation of distributed virtual en- MoMuSys (Mobile Multimedia Systems) is project number vironments.” (Eric Badiqué, Commission Project Officer.) AC098 in the ACTS (Advanced Communication Technology and Services) programme, funded up to 50 % from the European The MoMuSys project covers parts of all the areas mentioned in Commission. The rest is funded by the partners involved. The the above paragraph. Through the close work with MPEG-4 we project started on 1 September 1995, and will run until autumn are contributing to the solution of relevant problems in all these 1998. The project leader (prime) is Dirk Lappe from Bosch. The areas. MoMuSys project has already produced good results, and is regarded by the Commission as a very successful project. 1.3 Area of work

1.1 Project purpose The boundaries between telecommunications, computers and TV/film industries are blurring. MoMuSys will address all three From the official project description: areas through the field trials. The multimedia coding technology will be developed and promoted through the MPEG stand- “The main objective of the MoMuSys proposal is the develop- ardisation work. ment of mobile multimedia technology for real use, flexible enough to achieve all functionalities that will be required in In addition to developing necessary technology for the new future multimedia applications. This will be performed by functionalities addressed by MPEG, MoMuSys will work on developing tools, algorithms and profiles which will be vali- developing the necessary network-related technology, particu- dated in usage trials on real time hardware. Close collabora- larly for mobile networks. tion with the MPEG-4 standardisation effort will ensure the international relevance of the work.” To validate the technical solutions developed, MoMuSys will carry out three specific user trials, where the latest technology In addition to this widely defined goal, one of the main reasons will be deployed, and assessments and feedback from the users for the EU to support the project is that it boosts the European will be reported. level of research and development and strengthens the industry in this very important area. 1.4 Project partners Developing new software-based solutions for multimedia communication is a very important part of the project. This work fits The project partners are as follows: very well with the rapidly increasing use of the Internet and of Organisation Short name Country Multimedia applications in general. The fact that MoMuSys is very active in the ISO/MPEG-4 development means that this Robert Bosch GmbH BOSCH Germany part of the work is linked to standardisation and is therefore Armines CMM France open for use outside the MoMuSys project. This may prove to be very beneficial to MoMuSys, the industry and users in gen- France Telecom CNET CNET France eral. The technology developed will not be hidden, but will be Centro Studi E Laboratori part of an open standard. Telecommunicazioni SPA CSELT Italy

1.2 EU ACTS initiative Teltec Ireland DCU TELTEC Ireland Deutsche Telecom AG FTZ DTAG Germany MoMuSys is intended to be a European initiative for pushing Fraunhofer Gesellschaft zur technical development and standardisation for multimedia. The Förderung der Angewandten ACTS programme provides the basis for forming a European Forschung E.V FHG Germany collaboration platform to influence international standardisation and to push the technical development in Europe. Heinrich Hertz Institut für Nachrichtentechnik Berlin From the ACTS sub-domain on Multimedia Content Manipula- GmbH HHI Germany tion and Management (MCM) newsletter we read he following: Instituto Suprior Tecnico IST Portugal “If the primary objective of the ACTS programme is to con- Royal PTT Nederland KPN duct R&D in the context of trials, it will also be strongly sup- Research KPN Netherlands porting leading-edge research aimed at developing the tools for the future multimedia landscape. Within ACTS, the pur- Laboratoires d’Electronique pose of the MCM domain will be to develop audio-visual sig- Philips LEP France nal representation, manipulation and management tools able Queen Mary and Westfield to support the creation of advanced digital multimedia ser- College QMW UK vices. More specifically, the activities of the group will cover multimedia signal analysis and coding with an emphasis on Siemens AG SAG Germany contributions to MPEG-4. It will also include synthetic and Telefonica Investigacion y natural hybrid coding (MPEG-4 SNHC group), syntax Desarrollo TID Spain (MPEG-4 MSDL group), 3D analysis, modelling, texture mapping and VR technology. Applications will range from 3D Universität Hannover TUH Germany

Telektronikk 3.1997 133 WG0 Project Management

WG1 Requirements

WG2 WG3 WG4 Algorithm design HW development of and Mobile Multimedia Field Trials Figure 1 Project structure of the verification Terminals MoMuSys project

Universidad Politecnica de WG3 Development of Mobile Catalunya UPC Spain Multimedia Terminals Alan Pearmain,QMW Universidad Politecnica WG3.1 Multimedia Terminal de Madrid UPM Spain Hardware Telenor AS Telenor Norway WG3.2 Multimedia Terminal Software Algorithms VTT Information Technology VTT Finland WG3.3 Integration and Test Centre Commun d’Etudes de Telediffusion et Tele- WG4 Field Trials Väinö Typpi, VTT communications CCETT France WG4.1 Field Trials with End The project structure is shown in Figure 1. User Groups WG4.2 End User Systems Design 1.5 Workgroups and workpackages and Implementation WG4.3 MoMuSys User The project is divided into workgroups and workpackages as Group Trials follows: WG0 Project management Dirk Lappe, Bosch 2 Technology areas WG1 Requirements Karlheinz Brandenburg, FhG MoMuSys is working to develop new ways of coding and communicating multimedia information. To achieve this, MoMuSys WP1.1 Internal requirements is working in several fields of technology. To be able to develop WP1.2 External requirements full solutions that can be tested on non-technical users it is necessary to go further than producing research results. We need to WG2 Algorithm development Liam Ward, Teltec develop actual solutions in software and hardware, after first developing the concepts and doing the system designs. WP2.1 Integration

WP2.2 Syntax and Control 2.1 Mobile telecommunication WP2.3 Network Adaptation MoMuSys is concerned with mobile multimedia communica- WP2.4 Video Algorithms tion. The project will consider several available technologies for WP2.5 Audio Algorithms mobile communications, networks for both high bitrates and low bitrates. WP2.6 SNHC WP2.7 Analysis and Coding The actual networks to be included in the trials are: at higher bitrates ISDN over DECT User trial with ordinary users in a test WP2.8 Audio channel coding, region in Finland error robustness and error GSM and PSTN Very low bitrate testing concealment for audio Wireless LAN High bitrate testing UMTS 3rd generation mobile communication.

134 Telektronikk 3.1997 Other networks may come into consideration during the project. 2.2.2 Video processing

Common to all these networks in the first two years is the fact The MoMuSys Video group is led by Ferran Marques from that we will be running the communication on the TCP/IP pro- UPC. tocol. In the third year we plan to use unprotected mobile channels and add our own error protection layers. Like audio the video coding activity in MoMuSys is closely linked with MPEG. The video coding method developed in 2.2 Signal processing MPEG was implemented in MoMuSys from a very early stage, and the MoMuSys implementation is one of the two official reference implementations of MPEG-4 video. This is an advantage Much of the activity is related to digital signal processing. This for MoMuSys, since it has resulted in a very strong position in is mainly done in the audio and video groups, which are de- MPEG. It also means that there is a burden on MoMuSys to veloping new processing techniques for audio and video. We always keep the software up-to-date with the latest changes also have an SNHC group, which works with Face and Body made at MPEG meetings. However, new proposals are often Animation. being developed by other companies on the MoMuSys software, and after acceptance in MPEG, the software is donated to 2.2.1 Audio processing MPEG and integrated by MoMuSys to become a part of the official reference software. The MoMuSys Audio group is led by Luca Cellario from CSELT. The texture coding is based on the ITU H.263 standard. One may say that this is the core of the method. Nevertheless, the In the audio group the work on developing new techniques is method has gone through a number of changes, and a very done through active participation in the MPEG audio group. important new dimension has been added, namely shape coding. The MoMuSys project is a driving force in this work, especially The core software based on H.263 was developed by Telenor through the active role of Bernhard Grill from FhG. earlier as part of the standardisation work for H.263, and donated to MoMuSys as background code. This software has been The MPEG-4 audio standard is in its current draft state a very very effective in promoting Telenor both in ITU, MPEG and in broad standard. It covers a wide range of bitrates, and also a MoMuSys. wide range of aural bandwidth. The current method consists of several algorithms: Shape coding implies that the video scene is actually automatically analysed to partition the scene into separate video objects. • AAC (Advanced Audio Coding). This is the high level algo- This process is called segmentation, and is very computationally rithm, covering bitrates from approx. 8 to 256 kbps, with the intensive. MPEG will not standardise a segmentation method, normal bitrate at 64 kbps. This algorithm is a time/frequency only the coded representation of the shapes. This means that it is method (transform method) with additional vector quantiza- very important to develop good tools for segmentation, since tion tools. they will not be provided through MPEG. There will probably • Broadband CELP (Code Excited Linear Prediction). This be an example implementation in the informative part of the method is a well known algorithm, and is here used for cod- standard. The MoMuSys partners are working actively on de- ing 7 kHz audio at bitrates mainly in the area 16 – 32 kbps. veloping such algorithms for segmentation. Many of the partners are also involved in the European project COST211ter, • Narrowband CELP. This is similar to the above method, but which is concerned with developing good techniques for seg- for coding speech at 3.1 kHz. The bitrates are normally mentation. around 8 – 32 kbps. • Parametric codec. This is an effective speech coder, with low After segmentation, MPEG standardises the coding of the shape bandwidth and low bitrate. The bitrates are typically from information, the texture and the motion information. The coding 2 – 8 kbps. method for the shape information was developed by Noel Brady from MoMuSys (Teltec) and Frank Bossen from EPFL. This list of parameters for the algorithms quickly gets outdated, since the proponents of the different algorithms are adding The video objects may actually be treated like individual features and extending both the bitrates and the aural bandwidth streams, and such streams may come from different locations covered. One idea with this system of several algorithms is that and be composed into the same scene. More about scene comit will be possible to change algorithm “on the fly”, that is, position below. when the bitrate changes, the audio coding method changes adaptively. In modern networks this is a very useful feature. 2.3 Multimedia applications The audio group is also working to develop a very exciting The application areas for MPEG-4 may be very diverse. From aspect of MPEG-4, namely the coding of individual objects. MPEG-1 and MPEG-2 we know that these standards were Audio objects may be coded separately, either after being sepa- targeted at special application areas. MPEG-1 was for digital rated by analysis or by separate recording. The audio objects storage (CD-ROMs) and MPEG-2 was mainly aimed at TV- may then be transmitted with different quality, bitrate or by dif- quality broadcasting, although the video part was also standard- ferent algorithms. This allows for new ways of interacting with ised by the ITU, and is thus suitable for two-way communica- audio material. tion (using the low delay tools).

Telektronikk 3.1997 135 In contrast to this, MPEG-4 is aiming at a much broader range an MPEG-4 decoder system. From the Binary Format for Scene of application areas. At the low end MPEG-4 covers the very Description (BIFS) the player builds up a scene graph, and an low bitrates and may be used for mobile audio-visual communi- executive module ensures that decoding, rendering and pre- cation, also two-way communication. For the high end MPEG-4 sentation is being done in a synchronised way. The compositor has added the same interlaced tools as MPEG-2 and may be module builds up and controls the scene graph, and takes care used for the same broadcast type applications. Recently, MPEG that all rendering and presentation is done. Figure 2 illustrates has also been adding a separate still image algorithm. the architecture.

In the MoMuSys project we are mainly concerned with mobile Figure 2 shows how the elementary streams from the multiplex communication, and therefore the work has been focused on the are connected to nodes in the scene graph. These nodes are low bitrates. Still, MoMuSys has added an extension to the pro- actual decoders, and the output is treated by the compositor, ject for working on the higher bitrates. which carries out the rendering of the individual sources and presents the data to the user. The composition information One of the main features that distinguishes the MPEG-4 from which describes the placement of audio, video and other sources other standards is the object based coding of audio-visual sig- in the scene is transmitted in a compact description (BIFS). nals. The system work is closely related to MPEG-4 (development 2.4 System communication and composition group “im1”). The successful 3D based player made by MoMuSys will be merged with a 2D player made by CSELT, and will be the official reference software for MPEG-4 Systems. The MoMuSys Systems group is led by Robert Danielsen from The 3D compositor was developed by Karl Anders Øygard from Telenor. Telenor. In the period July – November four people from MoMuSys have direct responsibility for developing parts of the Apart from the coding of audio and video signals (both natural MPEG-4 system software (3D/2D Compositor, Multiplex, and synthetic), the system design of MPEG-4 is a very exciting BIFS, Audio composition). development. The MPEG-4 System specification will probably prove to be the most important part of MPEG. Even here The MoMuSys Systems group will build further on this plat- MoMuSys has assumed a central position. Olivier Avaro from form and add parts that are necessary for our field trials. Some MoMuSys (CNET) is the leader of the MPEG Systems group, of these parts may in turn be contributed to MPEG. and MoMuSys has contributed considerably to the common reference software development in the group. 2.5 Combining natural and synthetic data The main features of the system is reflected in the architecture, where elementary streams are coming from a demultiplexer into One of the new areas in the MPEG standardisation work is covered in the SNHC group (Synthetic Natural Hybrid Coding), where synthetic data is combined with natural data. The background for this work is that synthetic data are not very efficiently coded with coding methods developed for natural data and vice versa.

Some work in this area is being carried out in MoMuSys, in workpackage 2SNHC. However, the work has been BIFS Compositor mostly related to the system level composition, using FAP (Facial Animation Parameters) to build animated faces into scenes. This kind of work is also done in Video decoder workpackage 2.1 (Integration).

In the MPEG system software development, Miralab Audio decoder from Geneva has made very good contributions to Face and Body Animation (FBA). This software is currently being used in the 3D compositor developed by Telenor Network FAP for MoMuSys. This was demonstrated at the 40th MPEG-meeting in Stockholm 21–25 July 1997. Demultiplexer FAP 2.6 Extensions to higher bitrate networks

The project has been extended to take into account the scaleable FAP coding for higher bitrate networks. The work will be focused on video coding. Both scaleable texture coding and scaleable shape coding will be worked on, and contributions will be made to FAP MPEG.

This work is to be done in WP2.7, and will also cover work on Figure 2 System architecture algorithmic parallelisation of coding tools.

136 Telektronikk 3.1997 2.7 Hardware-related development audio coding, and also in the field of multiplexing and mobile communication. This work is carried out in WG3, led by Alan Pearmain from QMW. 3.3 Other external bodies This group is responsible for specifying the appropriate hard- The project is actively participating in other standardisation ware terminal to be used for running the software solutions from bodies and consortia. People in the project also contribute to WG2 in the MoMuSys field trials. In the first phase terminal we conferences and workshops in the area of coding and communi- are using a DSP-board in a portable Pentium based machine. cation. All this contributes to the successful promotion of the For phase 2 we will include an additional DSP card for in- MoMuSys work. creased processing power.

Most of the partners will purchase identical systems, to follow 4 Results achieved to date the specification from WG3. We will then have a base of these systems, which can interoperate in various networks. The termi- The MoMuSys project has been a success and has contributed nal will also be used in other ACTS-projects, for Telenor we significantly to the development of standard solutions for the note the projects CRABS and SINUS. coding and transmission of multimedia information.

2.8 Field trials 4.1 Software

The MoMuSys field trials will be carried out in three phases. Software has been contributed to MPEG for Video coding, The trials will be done with “End users” and “Expert users”. Audio coding and Systems. This software will be donated to MPEG, and will become part of the actual MPEG-4 standard. The work with end users is conducted within the frames of two This will benefit the industry and all users of the standard. scenarios, “Traveller’s AV Information Services for Mobile MPEG is also considering interoperability with other standards, Users” and “Mobile Remote Security Surveillance”. The first and may prove to become a very flexible and useful standard. scenario is very mass market oriented in nature and, accordingly, is targeted to layman users. It mainly deals with AV 4.2 The MoMuSys terminal database access. The second one is targeted to professional users in the security service field dealing principally with real- The first phase MoMuSys terminal has been specified, and soft- time AV information transfer. ware has been developed and optimized for the chosen platform. Most of the partners have acquired the terminal and will use it 3 Strong links to international for testing and evaluation, and for demonstrations.

standardisation Work is underway for designing the next version of the termi- 3.1 MPEG nal, which will be ready in January 1998.

MPEG (Moving Pictures Expert Group) is a nickname for an 5 Project management in the EU ISO subgroup: The MoMuSys project has proved itself to be a success. It con- ISO/IEC JTC1/SC29/WG11 sists of 20 partners, a large number for a research project. The reasons for the success are probably several: ISO International Standardisation Organisation • Enthusiasm about the field of research. The MPEG-related IEC International Electrotechnical Commission work is very relevant for the continued developments in Mul- JTC1 Joint Technical Committee 1 (Information Technol- timedia-related fields like Internet, videoconferencing, broad- ogy) casting and digital storage. We will see other applications maturing in the years to come. MoMuSys and MPEG are SC29 Subcommittee 29 (Coding of audio, picture, multi- working to provide the foundation for further developments media and hypermedia information) in these areas. WG11 Coding of moving pictures and audio • The project management works very well, and the work is being followed up closely from the European Commission. The connection to the work in MPEG is very strong in There are frequent deliverable reports to be written, with well MoMuSys. The work in MPEG is very important for the stan- planned schedules. dardised development of coding methods for multimedia signals. The Commission also appreciates this, and is very keen to • The work in the project is highly relevant to manufacturers, ser- contribute to the MPEG work. vice providers and universities. We have a good mix of these in the project, and the co-operation is going surprisingly well. 3.2 ITU • The social side of the project is also not to be forgotten. At every meeting we arrange a sports or leisure activity, which The project is also contributing to work in the ITU (Inter- certainly also contribute to the informal and friendly atmo- national Telecommunication Union) in the fields of video and sphere in the project.

Telektronikk 3.1997 137 6 Continuing the work 7 Conclusions

6.1 Each partner to his own As seen from Telenor’s viewpoint, the work in the MoMuSys project is very satisfying. It provides a base for highly relevant All the partners will get a strong foundation for further develop- research and development, with a 3-year contract. This is very ments in the area of multimedia coding and communication. important for keeping up a sustained effort in this field of work. The partners are not only observing the industry and acting accordingly, we are actually shaping parts of this industry The project is very successful, receives wide acclaim, and has through our research work and contributions to standardisation. dedicated partners who have a genuine interest in ensuring the success. This project certainly shows that the ACTS programme 6.2 Strengthening European industry works very well in this area.

Several of the partners have internal product developments related to the MoMuSys work. The fact that MoMuSys is so active in standardisation, and is actually developing the reference implementations for the industry means that the partners are in a very good position to be first on the market with their new products.

6.3 Towards MPEG-4 extensions and MPEG-7

The discussions on extensions and of starting new projects have begun in the MoMuSys group. One of the very interesting new developments is to be seen in MPEG-7, which will be concerned with content description.

In view of the rapid deployment of the multimedia coding standards, there is an increasing amount of multimedia data existing on the Internet, in databases, with broadcasters, in live feeds, etc. To be able to find a specific sequence of video or data today one has to look at the data itself. MPEG-7 will provide a standardised solution for this, by tagging the contents with descriptions of the content. These descriptions may then be accessed by agents or search tools to find the needed information.

Robert Danielsen is Research Scientist at Telenor R&D. He has been working with video coding in international standardisation and projects for several years. Currently, his interests are in the development of standardised solutions for multimedia coding in MPEG-4, and the commercial exploitation of this work in Telenor. email address: [email protected]

138 Telektronikk 3.1997 The SINUS project, Satellite communication in the UMTS BY VENDELA PAXAL

The SINUS project, Satellite Integration into Networks for Table 1 SINUS consortium partners and resource allocations UMTS Services, is an ACTS project (AC212) that started on 1 July 1996, and has a planned lifetime of about two and a Status Partner Country MM allocation half years, with a total human resource allocation of 666 Man-Months (MM). The project’s consortium consists of Partner 1, prime Alcatel Espace F 150 nine partners, which are listed in Table 1 with the respective Associated partner, France Telecom CNET F 30 MM allocations. The main objective is to define techniques partner 1 for integration of the satellite component into the UMTS (Universal Mobile Telecommunication System), with special Partner 2 CSELT I 58 focus on the interworking with the terrestrial component. Partner 3 Matra Marconi Space UK 24 This interworking contains critical aspects such as intersegment handover, call handling and mobility management. In Partner 4 Bradford University UK 56 SINUS, several of these aspects will be studied and the de- Partner 5 Surrey University UK 62 veloped solutions will be demonstrated. The particular features of the mobile satellite systems, including Doppler Partner 6 Fundacion Airtel Movil E 28 frequency shifts, important propagation delay, low transmit Partner 7 Telenor R&D N 44 power budget and severe radiowave propagation environment, necessitate a careful design of the modem and the air Partner 8 VTT Electronics SF 73 interface. Hence, an equivalently important goal of the Sub-contractor, Elektrobit Ltd SF 11 SINUS project is to find solutions to these problems, and to partner 8 define reference models in order to propose a common European platform for meeting the challenge of such system Partner 9 SINTEF N 130 constraints. Further, participation in standardisation bodies Telecom & Informatics is viewed as part of the outcome of the SINUS project.

Three major phases of the SINUS project have been defined. The general system study and the SINUS test-bed specification are comprised in the first part, development of the testbed constitutes the second, and finally the demonstrations will be performed in the third part. In the SINUS test- trial solutions to a great extent. This will not be possible in all bed there will be a mobile terminal connected to a fixed cases, thus requiring some dual mode operation blocks in termi- earth station through a satellite channel simulator. The nals designed for operation in both satellite and terrestrial fixed earth station provides access to a core B-ISDN net- UMTS networks. work represented by the National Host ItalHost in Torino, and a transparent terrestrial emulator will ensure the sig- One of the consequences of the integration of the S-UMTS into nalling link necessary to demonstrate intersegment han- the T-UMTS, is the adoption of the GRAN (Generic Radio dover algorithms. The SINUS project is today about to fin- Access Network) principle for describing the system architec- ish the first phase, and is thus entering the second. Co-oper- ture in the SINUS project. This GRAN principle has been de- ation with other ACTS projects has been initiated, such as scribed in [1] and [2], and for SINUS, the main interest in this co-operation concerning the radio access interface with concept at the actual stage of the project, is the division of the FRAMES, similar system architecture descriptions and pro- architecture into radio access dependent and radio access in- tocol definitions with RAINBOW, and finally an agreement dependent parts. In this manner the architecture description for using the multimedia terminal developed by MOMUSYS approach used in the SINUS project is compatible with the ITU as an application for the SINUS demonstrations. recommendations and the other European projects working on different aspects of the UMTS. The protocol layer description and the application of the GRAN principle to this description 1 Introduction are developed in section 3 dedicated to the SINUS test-bed presentation. We will in section 2 concentrate on the satellite sys- The goal of the SINUS demonstrations is to show the feasibility tem particularities and the critical aspects of the mobile satellite of an efficient integration of the satellite component into the communication and the integration into the terrestrial compo- terrestrial cellular UMTS network. As a consequence, the prob- nent. lems related to this integration will be identified, studied and adequate solutions will be proposed. The terrestrial approach to UMTS is different from what is natural from a satellite oriented 2 SINUS demonstrations point of view, which is due to the different types of radio propagation environment, with their restrictions and possibilities. It was natural to start the study in SINUS by identifying the par- Concerning the integration of one component into the other, it is ticular problems encountered on a mobile satellite link, and then a goal to achieve as high similarity between the technical solu- turn towards the difficulties of an integration into the T-UMTS. tions in the two systems as possible. Hence, this becomes an The radio propagation environment for mobile satellite commu- extra challenge to the system developer, since solutions found nication is characterised by rapid variations due to the move- for the terrestrial system are not always optimal in a satellite ment of both the satellite and the terminal. The evolution of system and vice versa. In Europe, the development of a terres- these variations depends on the satellite constellation earth orbit trial solution is in advance on a satellite solution, which means and the environment in which the terminal moves. In SINUS the that the development of S-UMTS will have to follow the terres- LEO, MEO and GEO constellations have been selected for

Telektronikk 3.1997 139 Table 2 Characteristics for candidate satellite constellations delay, required bandwidth and carrier frequency, number of users and traffic burstiness to mention only a few. This is why many projects propose the use of hybrid schemes, thus meeting Characteristics LEO MEO GEO Units/ both different types of impairments in a quasi-optimal way, and Notes also different types of system designs for multimode connec- Orbit height 700–4000 10000–25000 36000 km tions, at the expense of a higher complexity. This problem has been of major concern to the SINUS consortium, and much Propagation 5-25 80-120 280 ms effort has been put into the co-ordination with other projects, in delay particular those describing the terrestrial solution, in order to Satellite Every Every Unlikely Typical achieve as much commonality as possible on this issue. The handover 10 min. 2 hours values final choice in SINUS will be oriented towards a CDMA solu- during call tion. Beam Every Every May Typical The integration of the satellite component into the terrestrial handover min. 10 min. occur values network requires the development of powerful network manage- during call ment algorithms such as mobility management, call set-up and Nb of satellites, >48 10-15 3 release and handover. The desire to keep the terminal design at near global a reasonable complexity level imposes that most of the network coverage management evaluation, decision and execution are performed by the network and not the terminal. In SINUS, the study of these functionalities has been one of the major tasks in the first phase of the project. In the project definition phase, it was decided to demonstrate the call set-up and release and the major aspects of handover, but not the mobility management due to limitations of the test-bed set-up (limited number of transmission channels and terminals). The mobility management has study (the HEO constellation has been omitted because of however been studied in the first phase of the project, and the coverage limitations and important delay). The characteristics of functionality covers both the identification of the geographical the different orbits are resumed in Table 2, where we also find location and the identification of the network to which the user an indication of the passing time. The environments in which is connected. Procedures such as registration/deregistration and the terminals should function are; rural, suburban, and urban, location update will be initiated by the mobile terminal, whereas the indoor situation will be studied during experimentation but the connection set-up procedure will be handled by the network. we know that this type of communication may prove difficult The call set-up and release functionalities are grouped into two and even impossible. The terminal types will englobe the hand- main parts, the session set-up/release and the call set-up/release. held, the vehicular, the transportable and the fixed installation. The session set-up comprises four actions which are the service It is clear that even within the mobile satellite communication provider selection, the access control verification, the set-up of area defined here, huge variations will be found between scenar- the terminal connection and finally of the user connection. For ios made out of different associations of elements from the three the call set-up, the called user is located, the service profile of categories satellite constellation, radio propagation environment the called user is verified together with the domain and terminal and terminal type. capabilities, then follows the paging, the terminating session and the bearer connection. The paging informs the called user The impairments met in a satellite channel are principally the that there is an incoming call, and if the called terminal is shadowing or blockage, the Doppler carrier frequency shift due mobile, the network initiates a terminating session before the to rapid satellite motion, high transmission delays for some con- bearer service is connected. stellation types, interference (from other systems, other channels, other users ...) and the low power budgets. Satellite com- Two types of handover are required for the integration of a munication has always suffered from restrictive use of available mobile satellite communication system into the terrestrial; inter- power due to the satellite component power supply. In mobile and intra-segment handover. The inter-segment handover con- satellite communications, we also meet power restrictions due cerns the handover from the satellite component to the terres- to the use of for instance a handheld terminal (not only due to trial or vice versa, whereas the intra-segment handover is related battery limitations but also for medical reasons.) These power to the switching between two spot-beams or satellites during a restrictions will amplify the shadowing and blocking problem. communication. The handover strategies selected depend on the The Doppler carrier frequency shift may be compensated to amount of processing considered reasonable in the mobile ter- some extent, but will always necessitate a robust carrier fre- minal together with the handover performance. Some of the quency tracking at the receiver. The transmission delay will requirements in UMTS are that the handover should in principle limit real-time, interactive applications. Signal processing algo- be seamless in order not to degrade the QoS, the number of han- rithms depending on information from the remote terminal, such dovers should be kept to a minimum and handover between dif- as for instance power control, will also prove sensitive to trans- ferent operators and in different environments should be possi- mission delay. Different types of interference constitute a com- ble. The compromise between these requirements and the plex problem, and much effort has been made over the past implementation complexity considerations leads to the selection years in order to determine the best access method to meet these of a backward strategy associated with a hard mobile assisted problems. No unique answer seems to have been given, since all handover with signalling diversity. Hard handover will lead to methods have their particular advantages and disadvantages and an interruption in the connection, but the duration of this inter- they are dependent on system parameters such as transmission ruption will be reduced by the signalling diversity technique.

140 Telektronikk 3.1997 Transmission Application layer containing independent Call control, Mobility Management Transmission Radio resource Layer 4 dependent

Signalling network layer Layer 3

Link access control layer Layer 2 Medium access control layer

Physical layer Layer 1 Figure 1 SINUS test-bed protocol layers with radio dependent and independent parts

The signalling diversity implies that a connection is established independent. The MAC contains functionalities such as radio through the new link before the hard handover is performed. resource management and link quality control, the LAC is con- The network will be the “brain” behind the initiation and execu- cerned with for instance the logical link call set-up and release tion of the handover algorithms, only assisted by the mobile ter- functionalities. The network layer is considered independent of minal in order to reduce the complexity in this unit. the radio access, so is the application layer with the exception of some radio resource management functionalities which are The type of services selected for the SINUS demonstrations are present also at this level. offered through a multimedia application. The choice was motivated by the possibility of demonstrating different services such Figure 2 shows the test-bed for the SINUS demonstrations. This as speech, video and data transfer, or a combination of any of test-bed consists of four main parts which are the mobile termi- these, all from the same terminal. The environment and type of nal (MT), the fixed earth station (FES), the satellite channel satellite constellation are expected to have varying impacts on emulator and the terrestrial channel simulator. The MT and the the QoS of the different types of services. FES contain many equivalent entities, such as the emulator interface (which will be implemented in IF for simplicity) the modem and the baseline. Roughly speaking, the interface and 3 SINUS test-bed description the modem contain the functionality of the physical protocol layer, and the baseline those of the access control layer. On the As mentioned in section 1, the SINUS test-bed architecture is earth station side, there is a connection to the National Host based on the GRAN (Generic Radio Access Network) principle. (NH), ItalHost in Torino. This NH includes an ATM-laboratory, This implies that the radio interface will be compatible with a thus necessitating a B-ISDN interface between the FES and the variety of core networks (GSM, N/B-ISDN, ...), all included in a NH. The ATM switch is connected to a core network as well as UMTS Generic Core Network. Figure 1 represents the protocol terminal equipment for user application interfaces. The SINUS layers defined for the SINUS test-bed. The four protocol layers project will add an intelligent network (IN) unity, which will are the application layer, the signalling network layer, the control the network signalling and the management from the access control layer and the physical layer. One of the main pur- NH. The transmission channel of main interest in the SINUS poses of the GRAN approach is to identify the radio dependent project is the satellite channel. Since the SINUS consortium will and radio independent parts of the architecture. This permits a not have any access to operating mobile satellite systems for the separation of the parts of the terminal which are interchangeable trials, an emulator has been procured as a replacement. This for any type of access network with those having to be modified emulator will contain two full duplex channels, which will be according to the radio access network the terminal will be con- configurable before and during simulation runs, in order to rep- nected to, and thus making the terminal design modular and as resent a realistic mobile satellite channel. The presence of two universal as possible. In Figure 1 we see that the radio de- channels will also enable simulation of configurations including pendent parts are concentrated on the lower network layers two beams or two satellites, essential for intra-segment han- (shadowed in the figure). The physical layer is obviously radio dover execution tests. The inter-segment handover tests are the dependent, and so is part of the access control layer. Layer two reason for the presence of a terrestrial segment simulator in the has been divided into two sub-layers, and the frontier of the SINUS test-bed. This simulator will only convey signalling radio dependency passes between those sub-layers. The sub- information from one side to the other in order to show the algo- layer called medium access control (MAC) is dependent on the rithmic validity of the inter-segment handover protocols. The radio access, whereas the link access control (LAC) is radio

Telektronikk 3.1997 141 Satellite Satellite emulator channel emulator

Emulator Emulator

interface interface FES

Modulator Modulator Interface Demodulator Demodulator simulator

B-ISDN ATM Baseline Baseline interface switch

Terminal National Host

Network Signalling interface BTS CSS Applications MSCP Data & users Service & Stream application Terrestrial segment simulator Figure 2 Schematic representation of the SINUS test-bed

terrestrial simulator will therefore be a transparent channel, i.e. tions by profiting from the experience and equipment develop- it will not represent the terrestrial channel characteristics. The ment of other projects. Some ideas on joint trial possibilities with application data stream will not pass this terrestrial channel sim- other ACTS projects have been presented both on short term ulator at all, but be directly connected between the application within the actual lifetime of the project, and also as possibilities terminals with a switch allowing the turn off and on of the data for future co-operations in an eventual extension of the con- stream according to the signalling information. On the mobile cerned projects. Information has been exchanged with the terminal side, the connection with the terrestrial simulator is FRAMES and the RAINBOW projects in order to harmonise the established through a network interface. This interface will radio access mode and the implementation of the GRAN princi- ensure the turnover of the link from the satellite channel to the ple respectively. Joint trials are planned with the MOMUSYS terrestrial channel at the command of the handover procedure (cfr. article in this magazine issue) and TOMAS projects. protocol. The handover execution assistance required in the MOMUSYS can provide a prototype multimedia terminal for the chosen inter-segment handover scheme, is provided by this SINUS trials, and the MOMUSYS consortium is also interested entity. The test-bed will be designed for user bit rates up to 64 in the possibilities of multimedia transmission over the satellite kbit/s, and the multimedia application proposed for the trials channel. The TOMAS project will have access to an operating will be connected to the SINUS test-bed equipment through an geostationary satellite, which will permit SINUS to test our con- ISDN interface. nection over at least one real satellite constellation, and TOMAS is interested in trying out their terminal on our mobile satellite channel emulator. In an extension phase of the two projects, joint 4 Joint trials possibilities trials with RAINBOW seem attractive. The RAINBOW project is the equivalent to the SINUS project but for terrestrial UMTS, In line with the spirit of European projects such as ACTS, one of which means that by performing tests with both S- and T-UMTS the goals in the SINUS project is to contribute with research demonstrators we will cover most of the integrated system con- results and test equipment which may be of profit to other pro- figurations. The realisation of these joint trials may prove to be a jects within the program. In a similar manner, the SINUS con- complex task, this is why they are not considered possible within sortium would be most interested in extending the demonstra- either projects’ actual framework.

142 Telektronikk 3.1997 5 Conclusions and perspectives List of abbreviations ACTS Advanced Communications Technologies & The work within the SINUS project is about to conclude the Services specifications constituting the first phase of the project. The second phase will mainly be concerned with the development and ATM Asynchronous Transfer Mode integration of the test-bed entities. Finally, the last months of B-ISDN Broadband Integrated Services Digital Network 1998 will be dedicated to the SINUS demonstrations in Torino, Italy. CDMA Code Division Multiple Access FES Fixed Earth Station The SINUS project has been particularly concerned with the specificities of a satellite component for communication within FRAMES Future Radio Wideband Multiple Access Sys- the UMTS concept, but has also tried to achieve architectural tems commonalities with T-UMTS through the adoption of the GEO Geo-stationary Earth Orbit GRAN principle. The QoS requirements of S-UMTS have been specified in the first phase of the project, and the development GRAN Generic Radio Access Network following in the second phase is intended to meet those require- HEO High Elliptic Orbit ments. Particular attention has been given to the aspects concerning functionalities such as handover and connection estab- IN Intelligent Network lishment. Tests defined in the SINUS project will be performed, ITU International Telecommunication Union but co-operation with other projects permit interesting demonstrations, such as for instance testing of a newly developed mul- LAC Link Access Control timedia terminal from the MOMUSYS project. Many other per- LEO Low Earth Orbit spectives are open for future trials, both in the direction of real environment tests when mobile satellite systems are made avail- MAC Medium Access Control able, or in the direction of enhanced test-bed demonstrations MEO Medium Earth Orbit including other parts of a global system such as for instance emulation of the terrestrial channel. The main challenge of this MOMUSYS Mobile Multimedia Systems project is however to show the possibilities and limitations of an MT Mobile Terminal integrated S-UMTS component in order to contribute to the standardisation and development of a European concept which N-ISDN Narrowband Integrated Services Digital Net- may again influence the development on a global scale. work QoS Quality of Service References RAINBOW Radio Access Independent Broadband On Wire- less 1 ITU-8.1/157-E. Modularity for IMT-2000/FPLMTS, RCSG, 10 February 1997. SINUS Satellite Integration into Networks for UMTS Services 2 ETSI- 23.05. UMTS network principle, SMG. S-UMTS Satellite UMTS TOMAS Inter-trial Testbed of Mobile Applications for Satellite Communications T-UMTS Terrestrial UMTS UMTS Universal Mobile Telecommunication Systems

Vendela Paxal is Research Scientist in the Satellite & Radio Systems Unit at Telenor Research & Develop- ment, Kjeller email address: [email protected]

Telektronikk 3.1997 143