Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Project Number: IST-2000-25187 Project Title: TORRENT Deliverable Security*: PU CEC Deliverable Number: D1.2 Contractual Date of Delivery to the CEC: 31.7.2001 Actual Date of Delivery to the CEC: 18.9.2001 Title of Deliverable: Requirements for Service Providers, Network Operators, Manufacturers Work package contributing to the Deliverable: WP1 Type of Deliverable**: R Editor: Martin Potts Contributors: R. Phillips (Genuity), R. Tolstra (Tesion), E. Scharf, J. Griffiths, P. Hamer (QMUL), I. Borges, P. Rolo (PTIN), V. Apostolopoulou (OTEC), B. Martinez (Versaware), J. Rossebo, T. Olsen, H. Skaug, T. Konstali, T. Opperud, B. Haram (Telenor)

* Security: PU – Public, PP - Restricted to other programme participants (including the Commission Services) RE - Restricted to a group specified by the consortium (including the Commission Services) CO - Confidential, only for members of the consortium (including the Commission Services) ** Type: R - Report, P - Prototype, D - Demonstrator, O - Other

Abstract: The requirement for service providers, network operators and manufacturers is essentially to satisfy the user requirements as well as possible, in a fast and flexible manner, and for the least cost. The user requirements were identified in D1.1: “User Requirements”, as being that: all of the required services should be accessible, with the limitation that the quality may be affected according to the capabilities of the access network and the terminal. Such presentation adaptation should be automated for the user (ie. the user need not be aware of the underlying network). When a choice of access and/or core network technologies are available, the most suitable one will be (dynamically) chosen according to the user's instantaneous requirement for best quality, fastest response time, or lowest price (based on the current network conditions). Where no specific requirements are given, the Residential Gateway and Local Access Point will decide based on the service selected, the terminal, the status of the available networks, previous experience and personalisation profiles.

Keywords: Service providers, network operators, manufacturers, system architecture, access network technologies, terminals, residential gateway, home networks

______1 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

0. CONTENTS LIST

EXECUTIVE SUMMARY ...... 5

1. INTRODUCTION...... 6

2. ARCHITECTURAL FRAMEWORK...... 7

2.1 ARCHITECTURAL OPTIONS ...... 7 2.2 ARCHITECTURAL ISSUES FOR TORRENT ...... 11 3. SERVICE PROVIDERS, OPERATORS AND MANUFACTURERS REQUIREMENTS ...... 12

3.1 SERVICES, SERVICE COMPONENTS, AND THEIR CHARACTERISTICS...... 13 3.1.1 Telephony Services...... 14 3.1.2 Internet Services...... 15 3.1.3 Entertainment Services (interactive, real-time, non-interactive, non real-time)...... 16 3.1.4 Environmental systems (alarms, lights, heating, …)...... 19 3.2 ADDRESSING THE SERVICE REQUIREMENTS (THE OPERATOR AND MANUFACTURER VIEWPOINT) ...... 21 3.2.1 QoS ...... 21 4. TERMINALS...... 30

4.1 TELEPHONY TERMINALS...... 30 4.2 INTERNET TERMINALS...... 30 4.3 ENTERTAINMENT TERMINALS ...... 30 4.4 ENVIRONMENTAL TERMINALS ...... 31 5. HOME NETWORK PROTOCOLS, INTERFACES AND FUNCTIONALITIES...... 32

5.1 WIRED IN-HOME NETWORKS ...... 32 5.1.1 Ethernet...... 32 5.1.2 Copper pair...... 33 5.2 WIRELESS IN-HOME NETWORKS ...... 33 5.2.1 Bluetooth...... 34 5.2.2 IEEE 802.11b...... 34 5.2.3 HomeRF...... 34 5.2.4 DECT ...... 34 5.3 HOME NETWORK SECURITY ISSUES...... 34 5.3.1 Open and closed services...... 34 5.3.2 Resource sharing ...... 35 5.3.3 Policies and requirements...... 35 5.3.4 Anti-virus software...... 36 5.3.5 Suggested home network policy elements ...... 36 6. THE RESIDENTIAL GATEWAY...... 38

6.1 HIGH-LEVEL SYSTEM REQUIREMENTS ...... 38 6.1.1 Impact on Internet access ...... 38 6.1.2 Impact on home networking...... 39 6.1.3 Functional requirements...... 39 6.1.4 Basic interface configuration...... 41 6.1.5 Performance issues ...... 41 6.1.6 System requirements ...... 41 6.2 EXAMPLES OF POSSIBLE RG TYPES ...... 41 7. ACCESS NETWORK PROTOCOLS, INTERFACES AND FUNCTIONALITIES ...... 48 ______2 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

7.1 FIXED ACCESS NETWORKS...... 48 7.1.1 ISDN: Integrated Services Digital Network...... 48 7.1.2 XDSL Technologies (HDSL, SDSL, IDSL, VDSL, ADSL)...... 53 7.1.3 Powerline Communications (PLC) ...... 59 7.2 OPTICAL FIBRE (GLASS OR PLASTIC) ...... 63 7.2.1 ATM PON ...... 67 7.2.2 Ethernet PON...... 67 7.3 WIRELESS ACCESS NETWORKS ...... 69 7.3.1 Wireless Radio Access Networks ...... 69 7.3.2 Wireless Optical Networks...... 73 7.3.3 LMDS...... 77 7.4 WAVELENGTH DIVISION MULTIPLEXING (WDM)...... 80 7.4.1 WDM in a PON...... 81 7.4.2 WDM in access networks ...... 81 7.4.3 Application areas of WDM ...... 81 8. THE LOCAL ACCESS POINT...... 85

8.1 THE LAP HARDWARE ARCHITECTURE IN THE TORRENT PROJECT...... 85 8.2 THE LAP SOFTWARE ARCHITECTURE IN THE TORRENT PROJECT ...... 89 8.2.1 The TORRENT software structure in the first trials...... 90 8.2.2 The TORRENT software structure in the second trials...... 90 9. CORE NETWORK PROTOCOLS, INTERFACES AND FUNCTIONALITIES...... 92

9.1 CARRIER NETWORKS...... 92 9.2 CATV NETWORKS...... 93 9.2.1 Broadcasted TV...... 94 9.2.2 Voice over Cable (VoCable) ...... 94 10. MAPPING OF SERVICE REQUIREMENTS TO NETWORK RESOURCES...... 95

11. VALUE-ADDED FEATURES ...... 101

11.1 SUPPORT FOR ACCOUNTING...... 101 11.1.1 Introducing New Charging Schemes...... 102 11.1.2 Technical Implications of Using a Charging Scheme...... 103 11.1.3 Financial Implications of Using a Charging Scheme ...... 104 11.1.4 Systems integration ...... 105 11.2 SUPPORT FOR ADDING NEW SERVICE PROVIDER OFFERINGS...... 111 11.2.1 Firewall Services ...... 112 11.2.2 Adaptation of presentation / Content reformatting...... 113 11.2.3 Server based Games / Applications ...... 113 11.2.4 Web hosting...... 114 11.2.5 Caching...... 114 11.2.6 Load balancing / Content switching ...... 114 11.2.7 Datawarehousing...... 114 11.2.8 User store area ...... 115 11.2.9 Redundancy...... 115 11.2.10 Intrusion detection ...... 115 12. VALIDATION SCENARIOS AND CRITERIA ...... 116

12.1 EXAMPLES OF SERVICES THAT COULD BE VALIDATED...... 116 12.2 FEATURES TO BE VALIDATED...... 117 13. CONCLUSIONS...... 118 ______3 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

14. REFERENCES...... 119

15. ABBREVIATIONS ...... 120

______4 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Executive Summary

The high-level user requirements deduced by TORRENT can be expressed simply: all of the required services should be accessible, with the limitation that the quality may be affected according to the capabilities of the access network and the terminal. Such presentation adaptation should be automated for the user (ie. the user need not be aware of the underlying network). When a choice of access and/or core network technologies are available, the most suitable one will be (dynamically) chosen according to the user's instantaneous requirement for best quality, fastest response time, or lowest price (based on the current network conditions). Where no specific requirements are given, the Residential Gateway and Local Access Point will decide based on the service selected, the terminal, the status of the available networks, previous experience and personalisation profiles.

Whilst Deliverable D1.1: “User Requirements” listed and analysed these requirements in some detail, this document concentrates on how operators and manufacturers can satisfy these user needs through a variety of networking technologies. Those being considered by TORRENT include the most widespread existing ones, such as ISDN and CATV, and emerging deployments, such as xDSL, powerline communication, wireless optical and LMDS. The mapping of technologies to underlying physical infrastructure is also presented.

Furthermore, this Deliverable introduces the control and management software features that TORRENT will develop to exploit the fact that users may be able to choose from a number of different home-, access- and core- network technologies. This control and management software will enable services to be routed to the most appropriate network, according to the instantaneous QoS requirements of the user.

The overall requirements for service providers, network operators and manufacturers is to have an agreed system architecture that enables the user requirements to be met in an efficient manner. This architecture is described in section 2. In order to develop the architecture into a fully functioning system, the services (service components) must be specified (section 3) and the equipment and functionalities for the home, access and core networks characterised, interfaces defined, protocols identified, and mapping policies (service components to network resources) must be derived. Each of these items is documented in an explicit section (4-9).

Value-added features include accounting, the support of a firewall at the local exchange, and the easy integration of specific service provider offerings, adapted for the user environment (current access network, home network and terminal capabilities).

To validate the technique, the appropriate hardware and software will be prototyped for feasibility trials. Some first ideas are given at the end of this document.

______5 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

1. Introduction

This document has been produced by WP1: Architectural Framework. WP1 defines the overall framework for the project, in terms of identifying the QoS requirements of typical services, and then mapping these requirements to the network capabilities of a variety of home networks (Home RF, DECT, IEEE 802.11b, Bluetooth) and access networks (CATV, fibre, xDSL, powerline communications, ISDN, wireless optical, LMDS). This architectural framework is fundamental to the rest of the project. This framework defines the context within which the QoS negotiation and service provision will operate, and it determines how services must be defined so as to enable them to be prioritised on the access network, and later routed onto the most appropriate core network. The framework defined in this Deliverable will be re-examined throughout the software development process, and especially following the first experiments.

This document begins with an overview of the TORRENT architecture framework, which is based on the typical network environments that currently exist, but extended with hardware and software functionality to provide more flexibility in the usage of the underlying networks for meeting the instantaneous requirements of existing and emerging services. These underlying networks, the associated protocols, and their capabilities for conveying specific services are then analysed in detail. Anticipated terminals, and the newly identified devices (the Residential Gateway and the Local Access Point - introduced in the architecture section), are also described, both in terms of the hardware and software.

With reference to the architecture, the document is then structured following a logical progression from the end user terminal, through the home networks, the Residential Gateway, the access networks, the Local Access Point, and finally the core networks (sections 4-9).

At the end of the document, some conclusions are drawn regarding the scope of the development work in the project and the trials that will be made.

______6 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

2. Architectural Framework The architectural framework must be designed in such a way as to be able to satisfy the first sub-objective of the project: to produce a test-bed to assess different options for residential access networks. A key feature of such an architectural framework is its ability to map service characteristics to network performance parameters.

The proposed architectural framework comprises: • All aspects of home network technologies and their connection to the access network via the Residential Gateway (RG). • The Local Access Point (LAP), which groups telecommunications, media and computing technologies. • The communication between the user, the RG and the LAP. • A software architecture covering the basic communication requirements between these entities as well as with the core network.

2.1 Architectural Options The network architectures that are present today have mainly evolved from services that had to be delivered over dedicated architectures. This situation has produced a multi-network architecture as shown in Figure 2.1. Each of these networks is completely independent of the other and employs a different type of transport. There is also no overall policy manager for the different networks and services. TORRENT therefore assumes that the RG needs the capability to select from a number of available co- existing access networks, the one that is most appropriate at the time for carrying the service (according to, for example, the selected QoS). The design will be modular and (having several physical interfaces) represents a practical and evolutionary approach. This approach also allows the demonstration of the fullest flexibility and greatest degree of network negotiation and selection according to customer’s quality and cost requirements on the access domain. The goal for TORRENT is to be able to treat the – possibly many and varied - access and customer premise networks as a single resource, capable of servicing all the communication requirements of the customer. A single physical type of access is an extreme subset of this architecture, which – whilst technical feasible, possibly more cost effective for the user, and perfectly applicable in some situations – may not be realistic in others. For example, regulations may demand the powering of the end-user terminal from the local exchange, when the mains power fails. Other reasons for a user having several types of access networks are: reliability, pricing policy changes and technical interoperability issues. Indeed, many users today already have the opportunity to be connected to a multiple of network operators (eg. tele-communications operators and CATV) and have the choice of different services from the same operator (eg. analogue, ISDN, ADSL). The functions in the RG are simplified if only one physical access network is available (as shown in Figure 2.2), but some of the complexity is then shifted to the LAP. For design purposes, TORRENT accepts that several different types of access network will co-exist. In this environment, customers can freely subscribe and un-subscribe to the access networks of their choice, without having to make any changes to the RG. The management of the information requests will be under a policy management scheme that works with the RG of the customer and the LAP, who might be one of the service providers.

______7 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

The essential innovative feature of TORRENT is that it develops - and demonstrates through a test-bed – intelligence to enable each type of network to be exploited for the customers’ instantaneous QoS expectations.

Entertainment service provider or network operator local access point gateway (LAP) STB

coaxial cable access network cable Internet modem service provider or LAN gateway network operator no overall policy manager Telephony xDSL modem service provider or gateway network operator

NA copper pair access network switch & local P access point customer network access network (SW / LAP)

Figure 2.1: TORRENT Architecture (user has connectivity to parallel access networks)

Whilst a desirable network architecture goal for TORRENT might be a single access and customer premise network to service all the communication requirements of the customer as shown in Figure 2.2, it is not reasonable to consider that such an architecture will be widespread in the market within the next ten years. The goal of TORRENT is to develop a single architecture that will permit the customer to use his terminals independently of the access networks that are available to him. This entails the enabling of multiple networks to operate as one and the evolution, if that is required, into a single access and customer premises network.

______8 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

service provider or Usage policy: Core network / ISP selection: network operator - service access network mapping - resource knowledge - user preferences -service requirements

gateway

LAN

service provider or single NAP network operator access network local (from a user access point perspective) (LAP)

customer network access network service provider or network operator Figure 2.2: TORRENT Architecture (user has connectivity to a single access network only)

This converged network architecture presupposes that the Telephony service will be incorporated into the bundle of other services. Whilst this is possible there may be customer, regulatory and evolutionary issues that will keep it as a physically separate service. Also, it is not reasonable to assume that all customers will be able to receive the broadband access transport services due to their location and the cost of installation. It will be many years before optical fibre reaches the majority of the customer base. Figure 2.3 shows an architecture that provides the basic Telephony service in parallel to the broadband service, which could be over optical fibre or coaxial cable. The Telephony service is also shown providing a low data rate (48 kbits/s) and xDSL service. By keeping these options, an evolutionary approach can be taken.

______9 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Core network / ISP selection: - resource knowledge service provider or -service requirements network operator Usage policy: - service access network mapping - user preferences

LAP NA P

other access networks service provider or LA network operator N gateway

n,xnv,xnv,xnv,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc gateway

NA copper pair local n,xnv,xnv,xnv,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc Telephony P access network interface access point (LAP) Telephony service provider or customer network access network network operator

Figure 2.3: Hybrid Telephony Architecture (user has access to both copper pair and other networks)

There are a number of reasons why a user may decide to use more than one access network, including: • Reliability (access to more than one LAP, via multiple physical access networks increases the availability of access to services) • The pricing policies of the service providers may change. Having a multiple choice allows a fast reaction to such changes • Regulatory conditions currently require a telephone to be powered from the local exchange in cases of emergency (loss of mains power). Whilst a user may access broadband multimedia services over fibre or CATV, a copper pair connection may also be necessary, in order to satisfy the regulatory conditions • It may not be possible to access some services via one access network or service provider’s LAP, due to the physical capabilities of the access network, or missing inter-LAP agreements. • The majority of ADSL modems delivered to consumers are “best effort” modems (QoS mechanisms are not supported), optimised for Internet surfing. Hence, voice applications can be disturbed by other applications that are accessed simultaneously. This will cause unacceptable delays for the voice application. Hence, if a “best effort” modem is installed at the user, then it is desirable for the user to split the Telephony service onto an analogue subscriber line or an ISDN line. • Security aspects of the access and core networks: An ISDN line to a physically protected local exchange is less prone to attacks than cable or ADSL. In traditional PSTN/ISDN networks the network elements are physically secured and remote access is restricted via dial-up modems. In the Internet arena, it is almost impossible to rely on physical security as geographical distances can no longer be considered obstacles for attackers with respect to cost. The use of an IP address for subscriber identification instead of a physical line termination also increases the opportunities for attackers as it will be easier for the attacker

______10 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

to avoid discovery eg. via IP spoofing. Cable and powerline networks are more vulnerable to eavesdropping than traditional PSTN/ISDN networks.

2.2 Architectural Issues for TORRENT Each of the access systems in use has its own set of protocols. Since the goal of the TORRENT project is not to develop an architecture that employs all new processes, the project will have to consider which of these protocols offers the best solution for customer access. Whilst it is not possible for the TORRENT project to address all the architectural options involved with customer access, the following tables show the protocols that will be considered in the TORRENT project.

Access Network Technologies Analogue Low rate Cable Optical ISDN xDSL LMDS Powerline voice band digital modem modem Physical Layer Coaxial Optical Copper pair Wireless Mains cable fibre cable

New protocol Adaptation to existing developed within protocols TORRENT

Table 2.1: Access Network Technologies considered by TORRENT

______11 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

3. Service providers, Operators and Manufacturers Requirements The following figure is expanded from D1.1 and explains the manner in which the data from an application is encoded and encapsulated for transportation over various networks.

Fixed MUX Raw audio Xfer mode Bearer W/FDM Optical TDM fibre Raw video (SDM) Coax

Stat MUX Copper pair (ATM) (XDSC) Data I/P (MPLS)

Encoding (and possibly compression)

Figure 3.1: Services and their Transportation

The following table summarises those characteristics / requirements from Deliverable D1.1 that are important for operators and service providers.

User Requirements Operator/Manufacturer Challenges (from D1.1) Personalisation To customise services to users’ needs Self Provisioning To give users control Flexible Billing To enable the move away from traditional billing paradigms Mobility To provide access to the same services wherever the user is located Speed To enable applications on demand, and always-on connections Quality Providing consistent performance

Table 3.1: Mapping User Requirements to Operator/Manufacturer Challenges

In order to develop solutions for the user requirements in Table 3.1 into a fully functioning system, based on the architectures in section 2, the services (service components) must be specified, the equipment and functionalities for the home, access and core networks characterised, interfaces defined, protocols identified, or developed, and mapping policies (service components to network resources) must be derived.

______12 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Consideration must also be taken of the transportation means in Table 3.1. These aspects are described in the following sub-sections.

3.1 Services, Service Components, and their Characteristics The required services and respective access and core networks are described in the TORRENT Deliverable D1.1: "User Requirements". In that document, the services are divided into four categories as follows: • Telephony • Internet • Entertainment • Environmental Network operators and service providers presently provide one or more of these services over dedicated access networks. This diversity leads to an overly complex network structure due to the different protocols employed in multiple layers of the protocol stack. There are a number of efforts to arrive at an integrated set of protocols, but these do not involve any test bed evaluation or focus on only on one or two of the services. One of the difficult parts of the TORRENT project is how to address the multifaceted Telephony service. This service has a legacy in customer culture, regulations and invested equipment. It also is viewed by some as an outdated service that will be replaced by Internet services. Others see that IP telephony can be bundled with other services as a “best effort” service co-existing with a more secure and reliable connection oriented Telephony service. The other three service areas are much more straightforward without the built-in problems of the Telephony service.

Figure 3.2: Services and their Delivery Mechanisms

The major architectural difference between the Telephony service and the Internet service, as discussed in the TORRENT D1.1 Report, "User Requirements," is where the intelligence that processes the information is ______13 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers located. For Telephony services it is located in a central office facility, but for the Internet it is in the customer terminal as shown in Figures 3.1 and 3.2.

3.1.1 Telephony Services The Telephony service presents a challenge for the TORRENT project, because it is viewed differently by each administration and operator or service provider. The legacy in customer culture, regulations and invested equipment means that certain features need to be retained in the basic architecture and their evolution has to be considered. The basic voice service has become a part of the global culture, and many of its attributes have become part of everyday life. These attributes are: • High system reliability • Guaranteed QoS (through the use of dedicated circuits and a signalling scheme, prior to connecting the call to check whether or not sufficient end-to-end capacity is available) • The ability of the system to identify the calling location (eg. for emergency services) • Remote power • Worldwide interoperability. • No customer involvement in basic service compatibility

3.1.1.1. Voice Service The present Telephony service has been focused mostly on voice, with support for dial-up modems. ISDN has found a role in the support of higher data rate access for Internet service, and the availability of 2 channels enables simultaneous voice and data communication. However, telephony needs to move forward if it is to support multimedia services in the same manner as it supports voice. This means that any new architecture needs to address customer devices such as screenphones and soft switches interconnected over copper pairs.

The Telephony services identified in the TORRENT D1.1 Report, "User Requirements" are voice, data and multimedia. The voice service should evolve into a digital one that is compatible with both multimedia and Internet voice services like H.323 over copper pairs. The remote power issue should be considered and an approach recommended. A low to medium (48 to 128 Kbits/s) speed data service should continued to be part of the telephony offerings, but it should not use the assets of the voice service as it presently does. This additional data capacity could be shared with the multimedia service.

System Intelligence

Subscribers Analogue voice Digital Switch and Control Analogue POTS Line Matrix Trunk Analogue POTS cards cards Back Bone Analogue POTS

Figure 3.3: Present Day Switched Analogue Voice Service (Intelligence is in the central office, the telephone is a "networked computer")

______14 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

The evolution of the analogue voice terminals, through digital voice terminals, is predicted to lead to simple “Multimedia Phones”, using the capabilities offered by xDSL (or even ISDN).

3.1.1.2. Multimedia Service

Customers System Intelligence

n,xnv,xnv,xnv ,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc Digital xDSL and Control n,xnv,xnv,xnv ,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc MM Phone n,xnv,xnv,xnv,xnvnxccnxv,xnx ,cn,cxvnx,zcvnxc n, xnv,xnv,xnv,xnvnxccnxv ,xnx,cn,cxvnx,zcvnxc n,xnv,xnv,xnv ,xnvnxccnxv,xnx n,xnv ,cn,cxvnx,zcvnxc ,xnv,xnv,xnvnxccnxv n,xnv ,xnx,cn,cxvnx ,xnv,xnv,xnvnxccnxv ,zcvnxc ,xnx,cn n,xnv ,cxvnx,zcvnxc ,xnv,xnv,xnvnxccnxv ,xnx,cn ,cxvnx,zcvnxc n,xnv ,xnv,xnv,xnvnxccnxv ,xnx,cn,cxvnx ,zcvnxc Data MM Phone CO CPU Back Bone

Digital POTS

Figure 3.4: Simple Multimedia Service (Intelligence is in the central office, the terminal is a "networked computer")

The customer terminal for multimedia services (see Figure 3-4) should be: • low cost (about 200 Euros) • thin client • touch screen and voice entry • easily extendable (daisy-chaining) • capable of having a keyboard attached (no mouse is envisaged) • combined fixed and wireless

The Multimedia exchange that handles the traffic from such devices should: • use simple information formats (voice, display screen) and • not compete with the Internet and its Web services

3.1.2 Internet Services The "Internet service" began as Internet browsing, file transfer and e-mailing, but value-added e-commerce services are being incorporated, to exploit the basic functionality of data-oriented interactivity; eg.: home shopping/banking, travel booking, and - with the availability of higher bandwidths (thereby reducing delays) – voice.

The following services were described in D1.1: • Voice (Internet) services • Internet browsing (home shopping, travel, banking) • E-mailing

______15 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

• File transfer (working from home)

3.1.3 Entertainment Services (interactive, real-time, non-interactive, non real-time)

Macro Cell PCS Micro

Back Bone Switched Video ATM/IP ATM Services ATM ISP and IP Services IP Subscribers MUX Switch

n,xnv,xnv,xnv,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc Digital Switch

n,xnv,xnv,xnv,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc n,xnv,xnv,xnv,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc n,xnv,xnv,xnv,xnvnxccnxv,x nx,cn,cxvnx,zcvnxc DSU n,xnv,xnv,xnv,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc Router n,xnv,xnv,xnv,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc Data n,xnv,x nv,xnv,xnvnxccnxv , xnx,cn,cxv n,xnv,xnv,xnv,xnvnxccnxv,xnx,cn,cxvnx,zcvnxcnx,zcvnxc Fiber TV n,xnv,xnv,xnv,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc SW VDSL Loop CO Switch Trunk MUX CPU Matrix Trunk DSU Router Voice Cards Cards PC ADSL

n,xnv,xnv,xnv,xnvnxccnxv,xnx,cn,cxvnx,zcvnxc PC Modem Line Back Bone

kxvsdfsldfjsldfj fggh kxvsdfsldfjsldfjfggh kxvsdfsldfjsldfjfggh Cards kxvsdfsldfjsldfjfggh Voice kxvsdfsldfjsldfjfggh kxvsdfsldfjsldfjfggh Trunks Digital Analog POTS POTS Multi media Phone

Figure 3.5: Entertainment Services (PC, TV, Stereo, Switched Video, Internet (including games))

3.1.3.1 Interactive TV and games Interactive TV invites the users to interact with the programmes and advertising, making it possible to access more information related to the programmes they are currently watching, to participate in discussion forums, to select among different camera views or to buy products that are being advertised. The users increase their options to decide, and change from being passive spectators to interacting actors. Interactivity is seen as a potential opportunity for new revenue streams for content providers. Advanced set- top boxes enable interactivity, promising to transform dumb TV sets into instant film libraries, communication tools, shopping outlets, games and general entertainment centres. The return channel required for interactivity can use the same physical medium for television distribution (in the case of bi-directional CATV networks), or use a different network (such as PSTN) should the TV be broadcast over the air (including satellite). Interactive TV services are mainly asymmetric, as the bandwidth requirement for the upstream direction is usually much lower than for the downstream direction. This is because the information that flows from the customer premises to the head-end is essentially only to signal user actions, such as the selection of the interactive link, browsing, game movements, and additional requests for information.

______16 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Currently, interactive TV services and games are also being offered by non-traditional networks, like xDSL. However, the bandwidth required to transmit good quality video demands that the copper portion of the network to the user is short.

The various degrees of interactivity and real-time nature of entertainment services can be summarised in the following figure:

REAL TIME TV / Video Games HDTV Streaming Interactive TV VoD

NVoD

INTERACTIVITY

Figure 3.6: The Interactive and Real-time Nature of Entertainment Services

This category of services began with television, but is becoming an integrated set of services that require broadband - and, increasingly, interactive - access (Infotainment). While there are many services being proposed for this category, the main services for the user include some video component. The bandwidth requirements for entertainment services vary widely, depending upon what is being offered. These requirements range from a few kbit/s for the upstream control, to Mbit/s for the video streaming (generally downstream only) of high quality movie pictures. Variable Bit Rate (VBR) encoding helps to lower the video throughput, by at least a factor of 2, when a small packet loss is acceptable. The liberated bandwidth can be used (for example) to increase the video size or reduce the start intervals in NVoD systems. MPEG-1 is a compressed digital video format that was designed to provide VHS picture quality in the broadcasting field. The need for better picture quality (but using higher bit rates) led to the development of MPEG-2. MPEG-2 is the format of choice for High Definition Television (HDTV), Digital Television, in- flight entertainment, and Digital Versatile Disc (DVD). MPEG is versatile, flexible and provides superior quality for all kinds of applications. MPEG's dynamic bandwidth is very scalable. It allows a high compression rate to be applied to parts of frames where there is little video or audio information and a low compression rate where there is a lot of information. Efficient compression means each file can be as small as possible, enabling more content on (for example) a DVD without sacrificing playback quality. The examples here can be differentiated by their level of interactivity and real time demand; i.e. interactive real-time services (like games and interactive TV), and interactive, but non-real-time services (e.g. Near VoD). Broadcasted TV (including HDTV) is an example of a non-interactive, but real-time service.

______17 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

For VoD, there is only a small traffic volume in the upstream direction for the selection/control of the video service and a significantly larger one in the downstream direction for the video data itself. The available bandwidth in the downstream direction determines the achievable video quality. The available bandwidth in the downstream direction determines the achievable video quality. Usually, the video data is stored in a format that allows the sending of the same pictures with different resolutions depending upon the capabilities of the access network. DVD-like quality in resolution, frame rate and artefacts, requires a bandwidth of approximately 4 Mbits/s. This is a reduction by a factor of about 10 from the raw video data. The nature of VoD (VCR-like functionality) requires a stringent real-time response time. The throughput must also remain very constant to maintain the quality of the video data. Interactive games have in common a more or less stringent real-time characteristic, which depends upon the speed of the game. It may be difficult to support the most stringent bandwidth requirements of some types of game applications over all access networks, but the required bandwidth does depend upon the specific game, it’s design, and the generated traffic pattern. The amount of data exchanged between a client and the server is also heavily depending upon the type of the game, as there are (for example) text based and graphical adventures, strategy games, jump and run games and first person shooters to name only a few. Also some games are more asymmetric than others, with the main processing performed at the server, and only the details of the actions being sent from the players. There is much research presently on this topic. It is quite usual to also have in parallel some audio communication between the players of a game. This adds a small amount of bandwidth, but no other requirements.

NVoD alleviates the core network requirements in terms of bandwidth. Requests for the same movie within a period of time are grouped together and served as a single multicast stream, which usually requires delaying requests or satisfying them only at predefined time schedule As with VoD, also for NVoD there is only a small traffic volume in the upstream direction for the selection of the TV channel or video film, and a significantly larger one in the downstream direction for the video data itself. There are fewer real-time requirements than for VoD, since the user can merely accept to view at one of a number of pre-determined programme start times. A constant propagation delay is not a problem, but there is a need for a very constant throughput for the video data to keep the playback running and the quality constant as well. The real-time requirement in the core network can be relaxed if the system is designed to buffer the video data at the edge of the network for a few minutes (though this assumes that the user is sufficiently patient to select the desired video some time ahead of actually starting the playback. The extreme case is to download complete video films to the user premises e.g. during the day while he/she is at work, so that they are available for viewing in the evening. This requires large storage capacity at the user premises, and due to the long time shift, it is debatable whether the term NVoD is still appropriate for this scenario.

3.1.3.2 Application Service provider Application Service Provisioning (ASP) refers to a changing paradigm in current software distribution and management strategies. Traditional business models refer to customers (either residential or business) purchasing and administering their own applications, basically paying a one-time fee independent of the usage of this application, and yearly payments by the user to obtain updates and the support by the software manufacturer. ASP introduces a different business model by which a software manufacturer leases its software over a communication network (ie. the Internet) to a user or group of users. The users pay back a periodic fee that is dependent upon the actual usage of the application. Automatic updates as well as built-in maintenance are other added value features under this new trend. ______18 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

ASP relies basically on a client-server architecture running on a stable (in terms of QoS) communication network. The server houses the application and associated databases, sending to the customer replicas of the computer screens that would be viewed on a traditional desktop. Therefore, guaranteeing the QoS of the communication link between server and subscriber is a vital part of the success of this new business model. The general trend in the ASP model is that the communications link will be based on the Internet, mainly relying on IPv4 (and later IPv6) protocols. The main terms that will affect an adequate deployment of the ASP technologies are server loads and the capabilities of the communications links. Under the EoNEP concept, the TORRENT architecture supports both of these items, particularly by offering service providers the opportunity to house the applications in the Local Access Point, instead of in the core network. The server at the edge of the network acts as a secondary server to the user. Under this architecture, the QoS of the whole service is isolated from the communications link and particularly from the state of activity of the internal Internet, which is unpredictable and may offer no guarantee for a reliable service. Furthermore, response times from the network will reduce, and traffic over the Internet backbones will also be reduced; thus lowering the cost of the communications link.

3.1.4 Environmental systems (alarms, lights, heating, …) The networking of equipment in the home will be a major growth area in the new few years. Driven by the need to inter-work between telephones, televisions, VCRs and PCs, the complete range of consumer goods (including heating, lighting and alarms) will in the future have the capability to be connected to the communications network. This will permit the remote reading of meters for gas, oil, electricity, water, etc., to become commonplace. Environmental systems, including alarms, meter reading, control of lights, heating, etc. are vital systems in every home. Today most of these devices are either stand alone, or connected to a proprietary network. This will change in the future, more and more of these devices will be networked and the networks will be more standardised. Even though these systems generates low bit rates, and the requirements on the transmission system are rather low, these services are very complex to handle because of the many participants involved and the security requirements. On the whole 5 different types of participants are involved: • End User • Service Integrator • Service Provider • Network Operator • Manufacturers

End User In a building the different ECDs (Environment Control Devices) are networked in one or more different networks. There are mainly two types of ECDs: • sensors that collects data from the environment. Examples are meter readers, motion detectors, temperature sensors, cameras, remote controls, etc. • actuators that controls an actual equipment

______19 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

The different sensors generate data contiguously that has to be analysed locally by some kind of server functions. Based on this analyse several things may happen • the data is logged • one or more specific local actuators are invoked • a message is sent to the customers mobile, or other customer equipment • a service provider is alerted • a service integrator is alerted. The control and configuration of functions may be performed either locally or remote by the customer, by a service integrator, or by the service provider.

Service Integrator Service integrators do not need to be present in this concept, but such functions seem to be advantageous for both the end users and the service providers. The function of a service integrator is to be a common access point for an end user to different services, and perform functions that are necessary and common for all service providers. The service integrator may also perform operation and maintenance of user equipment necessary for providing these types of services.

Service Provider The service provider in this context is the provider of the actual end service to the end user. For example, an alarm company that offers alarm services to customers, and when receiving an alarm message, performs a predefined set of actions. There may be several service providers for each service, and a service provider may handle several different services.

Network Operator The role of the network operator is to convey this type of information at any time to any location in a secure, cost effective and reliable manner.

Manufacturers Manufacturers should develop equipment that is: • Connectable to a non proprietary network • Reliable and secure • Low cost • Easy to install, use, control and maintain.

3.1.4.1 Security The Service Provider and/or Service Integrator is responsible for providing the service with adequate security mechanisms in place. For example, if access to the service is provided over the Internet, then some form of strong authentication should be provided to prevent fraudulent use of the service or sabotage. Non- repudiation should also be required so that the abuser/offender can be held accountable for actions. However, this is one of the most difficult problems facing the security industry. This may be achieved using certificates. ______20 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Limiting access to the service to calls from only one mobile telephone number may be an acceptable option, although this solution has drawbacks.

3.2 Addressing the Service Requirements (The operator and manufacturer viewpoint) The ability truly to deliver quality of service will separate the winners from the losers in the packet-switched future. 1n broad terms, the quality of service of a wide-area network is a measure of how well it does its job-how quickly and reliably it transfers various kinds of data, including digitised voice and video traffic, from source to destination. Back when networks dealt pretty much exclusively with voice telephony, the subject hardly ever came up. The circuit-switched telephone system was designed specifically to satisfy the human ear. It did, and it does. Nowadays, with the advent of packet switching and the proliferation of many kinds of communications traffic (time-sensitive financial transactions, still images, large data files, voice, video, etc.), there are more than one set of criteria to satisfy. The data rate needed for satisfactory voice communication may take an intolerable time to transfer high-resolution images. Conversely, the degree of network latency acceptable in transferring some files may not be adequate for real-time voice. So QoS has become a hot topic, and the contracts that specify it, called Service Level Agreements (SLAs), are becoming more and more common, at least between service providers and their largest customers. In fact, as incumbent providers of telecommunications service are increasingly being challenged by competitive carriers, QoS has become a convenient marketing tool for both. The long-distance carrier AT&T Corp., for example, offers standard and gold versions of its SLAs. Rebates are credited to customer accounts when guaranteed service levels are not met.

3.2.1 QoS Technically, QoS refers to an aggregation of system performance metrics. The five most important of these are:

• Availability: Ideally, a network is available 100% of the time. Criteria are quite strict. Even so high- sounding a figure as 99.8% translates into about an hour; and a half of down time per month, which may be unacceptable to a large enterprise. Serious carriers strive for 99.9999% availability, which they refer to as "Six nines," and which translates into a downtime of 2.6 seconds a month.

• Throughput: This is the effective data transfer rate measured in bits per second. It is emphatically not the same as the maximum capacity, or wire speed, of the network, often erroneously called the network's bandwidth. Sharing a network lowers the throughput realisable by any user, as does the overhead imposed by the extra bits included in every packet for identification and other purposes. A minimum rate of throughput is usually guaranteed by a service provider.

• Packet loss: Network devices, like switches and routers, sometimes have to hold data packets in buffered queues when a link gets congested. If the link remains congested for too long, the buffered queues will overflow and data will be lost. The lost packets must be retransmitted, adding, of course, to the total transmission time. In a well-managed network, packet loss will typically be less than 1% averaged over, say, a month.

______21 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

• Delay: The time taken by data to travel from the source to the destination is known as delay. Unless satellites are involved, the latency of a 5000 kms voice call carried by a circuit-switched telephone network is about 25 ms. For the public Internet, a voice call may easily exceed 150 ms of delay because of: signal processing (digitising and compressing the analogue voice input) and congestion (queuing).

• Jitter (delay variation): This has many causes, including: variations in queue length; variations in the processing time needed to reorder packets that arrived out of order because they traveled over different paths; and variations in the processing time needed to reassemble packets that were segmented by the source before being transmitted.

Applications vary in their QoS requirements (see Table 3.2). A long file transfer needs a high throughput and low packet loss, but is not very sensitive to delay and jitter. Live videoconferencing, on the other hand, also needs high throughput, plus it is sensitive to both delay and jitter. It is these differences that must be considered in writing the SLAs between service providers and their clients. The usual agreement specifies the end-to-end performance to which the client is entitled over a specified time interval - a month or a quarter, for example.

Table 3.2: Applications and their QoS Requirements

QoS is largely about priorities. At network aggregation points, like routers, multiplexers, and switches, data streams with different QoS needs are combined for transport over a common infrastructure. Satisfactory QoS has two main requirements: a means for labelling flows with respect to their priorities, and network mechanisms for recognising the labels and acting on them. Some networks - notably, those that use the Asynchronous Transfer Mode (ATM) protocol - have extensive provisions of this kind.

______22 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

3.2.1.1 ATM ATM specifications define service classes according to the basic characteristics of the main communication applications and define many QoS parameters to specify the required service with a high granularity. ATM introduced among others new concepts like fast switching using a constant length of packets (cells), the virtual circuit (VC)/virtual path (VP) concept and the dynamic hierarchical organisation and routing using the Private Network Node Interface (PNNI) protocol. Partly because of it's complexity in terms of signalling, traffic management and resource allocation, and especially the existing predominance of the Internet Protocol for PC-based multimedia and data applications, ATM never succeeded to become the single protocol that would enable the convergence of networks and applications. Unfortunately, the lnternet does not, and neither do the similar IP networks based on the Transmission Control Protocol/ Internet protocol (TCP/IP) suite. IP is a best-effort protocol in that it does not guarantee delivery of data packets. Confirmation of the arrival of data packets at the destination is the responsibility of the TCP, which sits just above the IP in the well-known seven-layer open systems interconnection (OSI) reference model promulgated by the International Organisation for Standardisation (ISO), a worldwide federation of national standards bodies. If any packet is not delivered (as determined by checking the sequence numbers of packets at the destination), TCP requests a retransmission of the missing packet, thereby ensuring that all packets eventually get to the destination. This is effective, but slow. Therefore, TCP is generally used by applications that are not time- sensitive. Real-time applications cannot take advantage of TCP. Obviously, the time needed for keeping track of missing packets and re-transmitting them is not acceptable in such cases. So these applications rely on what is essentially a stripped-down version of TCP known as the User Datagram Protocol (UDP), which runs faster than TCP by omitting some of its functionality. Applications that run over UDP must either have those missing capabilities built into them or else do without. In the case of voice communications where re-transmitting packets takes too long to be of any value anyway, missing packets are simply lost. Internet telephony, therefore, will work only over networks that are quite reliable to begin with, like fibre-based networks with modern switches and routers.

The IETF - the protocol engineering and development arm of the Internet Society, has proposed several methods for improving QoS, including IntServ, DiffServ, and MPLS (see Figure 3.7). Some typical applications are indicated in the top layer of the diagram, while the second layer shows the different procedures proposed by the task force for handing them.

______23 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Figure 3.7: Classes of service and the mechanisms for supporting their QoS requirements

3.2.1.2 IntServ Integrated service (IntServ) is the earliest of these procedures. It assigns a specific flow of data to a traffic class, as it is called, which defines a certain level of service. It may, for example, require best-effort delivery. It might even impose some limits on delay. Once a class has been assigned to the data flow, a so-called path message is forwarded to the destination to determine whether the network has available the resources (transmission capacity, buffer space, etc.) needed to support that specific class of service. If all devices along the path are found capable of providing the required resources, the receiver generates "resv” message and returns it to the source indicating that the latter may start transmission of its data. The procedure, known as the resource reservation protocol (RSVP), is repeated continually to verify that the necessary resources remain available. If the required resources are not available, however, the receiver sends an RSVP error message to the transmitter. Although IntServ has some attractive aspects it does have its problems. One obviously, is that it has no means of ensuring that the necessary resources will be available when wanted. Another is that it reserves network ______24 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers resources on a per flow basis. If multiple flows from an aggregation point--say a communications sever in a local-area network--all require the same resources, the flows will nevertheless all be treated individually. The resv message must be sent separately for each flow. In other words, IntServ does not scale well, and so wastes network resources.

3.2.1.3 DiffServ The procedures of IntServ are improved upon in another method from the IETF, one known as differentiated service (DiffServ). With DiffServ, a short tag is appended to each packet depending on its service class. Data flows having the same resource requirements may then be aggregated on the basis of their tags when they arrive at the edge routers. The routes at the core can then forward the data flows toward their destinations on the basis of their tags without examining the individual packet headers in detail. Since most of the decision- making is in this way transferred from the core routes to the edge routes, the core network runs much faster In the past, QoS planners supported both IntServ and DiffServ. At present, however, the trend is to use DiffServ supplemented by some of the resource reservation capabilities of RSVP at the edges.

3.2.1.4 MPLS A newer approach to speeding the transmission of data through a network is Multiprotocol Label Switching (MPLS), also a procedure promulgated by the IETF. Normally under IP, packet headers are examined at every transit point (multiplexer router or switch) in a network. This takes time and contributes to the overall data delay. A more efficient approach would be to label the packets in such a way as to make it unnecessary for each IP packet header to be analysed at points intermediate between the source and destination. MPLS does this by appropriately labelling IP packets at the input of label edge routers located at the entry points of an MPLS-enabled network (see Figure 3.8).

Figure 3.8: MPLS

______25 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

The procedure works as follows: the label edge router examines the incoming packets and decides - based on the packet’s source address, destination address, and priority level - where to send it for its next hop through the network It also attaches a 32-bit tag, known as an MPLS label, to the packet. The MPLS label contains such information as whether the packet should be treated as MPLS traffic or routed as an ordinary IP packet; whether it conforms to IPv4 or IPv6; the packets “time to live”; and, of course, what its next hop should be The edge router then forwards the packet to the router at the end of the next hop. That router, in turn, examines the MPLS label and decides on the next hop for the packet. That second router then creates a second MPLS label. The two labels are swapped before the packet is forwarded to the second hop. The process is repeated until the packet reaches its destination. This procedure has two advantages over normal IP routing: • the routers along the path need not read and analyse a packet’s complete header information, just the shorter MPLS label. This alone saves some time. • the swapping of labels leaves a trail in the registry of the routers that other packets in the same session can follow. Once the packet establishes a path, decision-making at intermediate points is eliminated to a great extent. This markedly speeds up the transfer of data. Many network service providers have installed label edge routers and are about to roll out MPLS services. Cable & Wireless has started offering MPLS for its transatlantic links, which join New York City and Washington, D.C., to London, Amsterdam, and Frankfurt, Germany. Cable & Wireless plans to introduce MPLS in all of its OC-192 (9.953 Gb/s) fibre networks between now and the end of 2001.

3.2.1.5 Common Open Policy Service Technologies that involve both software and hardware now exist to detect the requirements of each data flow on the fly - inferring them from, say its source or destination IP address instead of reading them from a special label. Once a specific application in a session is detected, it can be given the priority to which it is entitled. But until recently, a client’s network administrator had to inform the service provider about each and every change in the priorities of data generated by certain applications. As this process costs time and money, many clients have been discouraged from requisitioning the enhanced services in the first place. However, a client can add advanced services much more easily, thanks to a new tool for assuring the QoS of a network. Known as the Common Open Policy Service (COPS) protocol, the tool is more adaptable to a customer’s own requirements, allowing those requirements to vary with time of day, application, or even user session. The requirements and the rules for allocation of system resources known as policies are decided in advance. The objective is to specify a service in unequivocal terms and to allocate the resources required to deliver that service. Policy information is stored in a policy server from where it is shared with other network devices using COPS. The rules follow an IF, WHAT, WHEN, and THEN logic. A typical sequence of events could be: IF: The user belongs to the computer-aided design group # 003 and WHAT: the application of the design of a rocket engine and WHEN: the time is between 0800 and 1400 hours on Monday through Friday THEN: the user is entitled to: a service level S, that gives a throughput of X kb/s with an end-to-end latency of no more than Y ms.

______26 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

The service level could also specify other parameters, such as constant-bit-rate service. Once a user has put such a policy in place, it becomes easier for the client’s network administrator to configure and adapt the system to the company’s changing circumstances. The three principal elements of such a policy-based traffic management system are (see Figure 3.9): • policy creation and storage • interpretation, • enforcement

Figure 3.9: The Common Open Policy Service

When a data packet arrives at the input port of the enforcement device, the device first determines the classification of the data by some predefined criteria. Then, using the COPS protocol and the well-established simple network management protocol (SNMP), it checks with the policy interpreter as to the QoS to which the packet is entitled. The policy interpreter, in its turn, verifies the status of the data by pulling the policy

______27 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers rules using the lightweight directory access protocol (LDAP), a protocol commonly used for exchanging information among directory databases. With the help of the information thus retrieved, the interpreter determines what are called the rights of that particular data packet. On receipt of information on these rights, the enforcement device sends the packet, properly tagged onward to a router. If the same type of data, such as a request to a specific Web site, is found to be repeating often, the rules could be temporarily cached in the enforcement device itself. Vendors, such as Cisco Systems, Juniper Networks, Extreme Networks, and Nortel, are already shipping servers and routers that can run COPS. And start-up service providers like Yipes are creating dynamic QoS- on-demand environments, to provide capacity adjustable in increments that are as small as 1 Mbit/s for time- sensitive applications. But interoperability problems do exist when attempts are made to have products from different vendors work together. These problems will have to be solved before policy-based networking becomes ubiquitous. Nevertheless there is a growing worldwide adherence to COPS.

3.2.1.6 Measuring and Monitoring QoS Notwithstanding the methods used for assuring QoS in a virtual private network (VPN), measuring and displaying the parameters are vital. If customers cannot feel assured of getting the service they are paying for, how likely are they to continue paying? Fortunately, the TCP/IP protocols are also well suited to measurement of metrics like throughput, forwarding rate, and packet loss. Several vendors, such as Micromuse, Visual Networks, Netscout, Infovista, Sitara Networks, Netcom Systems, Lightspeed Systems, and CrossKeys Systems specialise in QoS monitoring, filtering, and reporting equipment.

3.2.1.7 Service Level Agreements (SLAs) For approaches, which do not apply an explicit signalling of traffic parameters, Service Level Agreements (SLAs) are specified between customer and operators. These SLAs again describe the allowed volume and characteristics the network input traffic has to keep to and probably also additional parameters like (for example) availability, but not on a per-flow basis. In both cases the network operator will ensure that the traffic corresponds to the contract or the SLA in order to be able to provide QoS at all and not to reduce the quality of other traffic flows by admitting ill behaving ones. The main difficulty in this context is always, that the QoS requirements and the traffic parameters have to be specified somehow. Currently, most applications are not able to do this and it is also very unlikely, that the customers will do this in a complicated way. Moreover, the customers are mainly interested in and want to have influence on their perceived quality that depends upon many factors. A mapping into the necessary parameters of a SLA or another traffic contract can therefore be demanding. This stimulates new approaches in the customer-operator interaction. The current trend caused by the availability of QoS in networks is the merging of classical telephony networks and data networks into converged networks which are able to support all types of services on one common networking technology. The need for an improvement of traffic control and network management in combination with better interoperation and ease of use is not only still there but moreover even reinforced because a degradation of already accustomed quality will not be accepted. This goal is precisely in-line with the aims of TORRENT. The SLA is where the provider’s technical competence, dedication to service, and business integrity is revealed. SLAs are “… one part technical, one part contracting, and three parts negotiating ….”. Carriers have a vested interest in minimising their exposure to penalties; end-users have an equally vested interest in

______28 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers maximising it. In order for both sides to maximise their gains, a clear understanding of one’s own business objectives is critical while drawing up a SLA.” A well-written SLA should include the metrics of QoS and factors such as availability, maintenance scheduling, and mean time to repair. The client has the right to know about the time needed for network recovery after a power outage or equipment failure. The client also should be aware of the provider’s ability to proactively detect and correct problems that may be looming. Automatic generation of QoS reports, alarms and trouble tickets, and issuance of credits for the vendor’s noncompliance should be an integral part of an SLA. Other key questions are: what type of QoS reports should be generated? and how often should QoS metrics be taken and reported? If a service provider’s report bases average availability on measurements taken over 24 hours, it may hide the problems that occur during the hours of peak usage. Generating billing and credit records as per SLAs have not yet reached a high degree of automation. Another challenge to delivering good QoS arises when a virtual private network crosses the administrative and technical domains of many providers, perhaps incumbent telcos and competitive providers, who may not all adhere to the same transmission and QoS technologies. Furthermore, while designing a wide area network for a high QoS, special attention has to be paid to the interfaces at aggregation points where there is a capacity mismatch between access links and network core links. Capacity mismatch occurs when a 100Mbit/s local-area network, for example, interfaces with a 1.5 / 2Mbit/s wide-area network line. The more diverse and important a client’s communications traffic becomes, the more crucial it is that the carrier maintain a high QoS. Throughput, availability, packet loss, latency, and jitter must all be spelled out in SLAs, along with how each is to be measured and reported. (It is not uncommon for carriers to track QoS but not report the results to the client unless an extra fee is paid. Also, don’t expect a carrier to generate credits automatically unless obliged to under the SLA.).

______29 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

4. Terminals Users have the following requirements that must be satisfied by in-house network providers and terminal equipment manufacturers: • All appropriate services should be accessible and controllable • Terminals should be inter-connectable • Easy to install (eg. wireless or PLC) • Easy to use, control, maintain • Ability to switch between different service types/qualities during a session • Global in use (same plug and protocol) = use of embedded systems = low cost • Reliable (self-monitoring) • Security of customer information and network resources

4.1 Telephony terminals There are two different schools of thought when it comes to telephony terminals. One is that the Personal Computer will assume this role, whilst there are others that want to keep the some type of telephone equipment like that is presently in use. There are a number of factors that will determine the direction the terminals will take; the two primary ones are customer demand and regulations. There are many people, at least for the next ten years, that do not want to have to use a computer for their primary voice communication needs. They are happy with just a telephone or some type of simple screen-phone. To eliminate the basic Telephony voice service as the basic option may not be acceptable. This could cause a regulatory impact. Another regulatory impact could be caused by the service rules that vary from area to area. The two main ones are "life line" service where the telephone has to work even if there is no customer power supply, and the emergency response service (“112” in Europe and “911” in the US) where the telephone number is used to locate the caller. Other items that might affect the overall architecture are related to tariffs. To mix the regulatory world of telephony with the market-based structure of the Internet will make a major change in the structure of IP networks. The alternative may be to have two separate services, Telephony and Internet, with their own customer cultures and regulations. However, these two services can coexist within the same access and customer premises networks. Figure 5.2 shows a possible terminal for Telephony services for both telephony voice and data.

4.2 Internet Terminals The basic Internet terminal today is the PC, and some form of this device, such as a palm pilot, will remain the main terminal for Internet services. The interfaces for these devices are well defined and TORRENT does not see that any changes are necessary.

4.3 Entertainment Terminals The entertainment terminals centre on two basic types. One is the traditional TV, accessed via a Set Top Box (STB) and the other is the Internet-based home entertainment complex and web TV that are appearing on the

______30 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers market. The main issues for the TORRENT project is the interface these will have to the CPN and the related protocol stack to permit them to be accessed by - or for them to access - another element in a remote network. This interface should not be complex and add very little to the cost of the entertainment terminal. However, the interface may involve two distinct formats - one for the data interface and the other for the video interface. The video interface will present the most challenges.

4.4 Environmental Terminals The environmental terminals will come in many forms that will differ greatly in functions and configurations. The main issues for the TORRENT project is the interface these will have to the CPN and the related protocol stack to permit them to be accessed or for them to access another element in a remote network. This interface should not be complex and add very little to the cost of the environmental terminal.

______31 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

5. Home Network Protocols, Interfaces and Functionalities

5.1 Wired in-home networks

5.1.1 Ethernet

SPLITTER

ETHERNET 10BASE-T ADSL ATMF 25 MODEMS USB

FILTER

Figure 5.1: The Home Network (ADSL-access)

The ADSL modem has an ADSL interface to the access network and 10Base-T Ethernet, ATMF-25 or USB to the client’s PC. The ADSL interface operates over one copper pair of wires. The only protocol over the ADSL, that is physical, is the ATM. The 10BASE-T interface operates over two pairs of wires, one pair used for receive data signals and the other pair used for transmit data signals. The two wires in each pair must be twisted together for the entire length of the segment, a standard technique used to improve the signal carrying characteristics of a wire pair. This interface terminates the ATM connections and extracts frames from arriving cells and encapsulates frames in departing cells. The ATMF-25 interface does not terminate ATM connections, it just switches ATM cells between the ADSL and ATMF-25 port. It is the ATMF-25 PC-NIC that actually initiates or terminates ATM channels. The ATMF-25 interface offers maximum TCP/IP transparency because it switches ATM cells and does not touch TCP/IP information. The Universal Serial Bus (USB) interface is a medium speed, plug-and-play technology found on most new computers. It outperforms interfaces like serial or parallel ports in terms of data transport capabilities. Devices are detected and configured automatically. Moreover, USB supports “hot swaps” or “hot plug & play”. This means that several USB devices can be connected and disconnected to the USB port without powering down the PC every time the user changes the device.

______32 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

5.1.2 Copper pair

analogue telephones n ,x nv ,x n v, xn v ,x nv nxc c nxv ,x nx ,c n ,c xvnx, zcv nxc master terminal copper pair

n ,x nv ,x n v, xnv ,x n vnxccnxv ,x nx ,c n ,c xvnx, zc vnxc

voice display display

copper pair with encoder driver modem

exchange power multimedia

multiplexer extension

low rate 3.4 kHz multimedia

modem voice data (digital)

PC/LAN interface (analogue)

D U local power PC with low rate low rate data xDSL exchange power Internet ADSL/VDSL connection

Figure 5.2: The Home Network (Copper-pair oriented)

In Figure 5.2, a copper-pair based home network installation is shown, which is also compatible with an ADSL access network. It enables voice and multimedia data over standard telephony wiring with: • The voice from the exchange modem unit as analogue voice to other telephones (existing ones) • The multimedia data service is carried on the same copper pair as the analogue voice as a modem signal • The primary voice and data services could be direct connections if the modem exchange unit is a module of the primary terminal unit

Some of the basic bandwidth (bit rate) could be used to provide a "cheap" Internet connection for the PC. The customer base will probably be divided into three groups for their communication needs: • Those that want only a PC and Internet service (includes Telephony service) • Those that want only Telephony services (voice or screen phone with enhanced features). • Those that want both services.

5.2 Wireless in-home networks Low price, acceptable performance and simplicity are the most important user requirements related to home networking. Wireless networks are flexible, and as the price drops to be close to the cost of a wired solution, this technology is expected to take the major share of the home network marked for moderate data rates (less than 100 Mbit/s). At present, four wireless technologies, DECT, IEEE 802.11b, HomeRF and Bluetooth compete in the marketplace for home wireless networking. In fact, Bluetooth does not compete with the two others, but rather supplements them, as Bluetooth is a personal network, or a cable substitute, operating at a low data rate with limited range. Also DECT seems not to be a competitor for wireless data, but is more applicable for local wireless voice applications. New systems as Hiperlan-II and IEEE 802.11a, with higher data rates, are under development. ______33 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

5.2.1 Bluetooth Bluetooth is a low cost, low power, robust wireless connection method with a small footprint that makes it very well suited for millions of handheld devices. Bluetooth operates in the license free 2.4GHz band and uses frequency hopping at a rate of 1600 hops/second. A Class 3 Bluetooth device emits 1 milliwatt of radio energy, consuming 50 to 150 milliwatts and the chip has a footprint of 8x8mm. The range is about 10 meters with a maximum speed of 721kbit/s for data. The price of a Bluetooth chipset, excluding application interface software, is expected to drop from $20 to $5 by 2003. Bluetooth is expected to have a reach this year with 20-30 million end user devices.

5.2.2 IEEE 802.11b IEEE 802.11b or Wi-Fi seems to be the winner of the present wireless home networking race. The system operates in the license free 2.4GHz band and provides a data rate of maximum 11 Mbit/s. The system has a range of about 300 meters outdoor and 30 – 100 meters indoor. Wi-Fi is backed by tech giants like Intel, Lucent Technologies, Cisco Systems, 3Com, D-Link and Apple Computer. Several products are available at affordable prices, and the price for the chipset is expected to fall to around $5 by 2003

5.2.3 HomeRF HomeRF is a wireless system operating in the 2.4 GHz band and support both synchronous and asynchronous data transfer. The current products are capable of a data rates at about 2 Mbit/s, but 11 Mbit/s products will be available in Q4 2001. Proxim and Motorola are makers of HomeRF-based products, but not many products are available.

5.2.4 DECT DECT is a digital wireless technology that originated in Europe, but is now being adopted increasingly worldwide, for cordless telephones, wireless offices and even wireless telephone lines to the home. Data rates at about 2 Mbit/s are possible.

5.3 Home network security issues The following describes some security aspects that will be addressed by the TORRENT project.

5.3.1 Open and closed services In a home network there may be users accessing open services eg. Internet surfing and there may also be users accessing closed services eg. home office services. Simultaneous access to open and closed services (eg. accessing the company network from a home office connection while accessing the Internet via an ISP) via the RG is a potential security risk for the closed service network. One can reduce the risk by making a policy that simultaneous access to open and closed services from one client PC must be prohibited. However, two clients accessing two different services from the same home network have a link between them - through the home LAN. One computer is accessing the Internet service while another computer is accessing the corporate network using the home office service. If the computer using the Internet service is compromised in some way, it can be used to attack the other computer, thus being a threat to resources of the closed service. In this case, the home network functions as a backdoor into the closed network service.

______34 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

5.3.2 Resource sharing Improper configuration of the file and printer sharing service of the Windows operating system represents a security threat. The Microsoft networking client offers the possibility to bind the file and printer sharing service to the TCP/IP protocol. This means that one can access shared resources on computers not only on the same LAN, but also across the entire Internet, and visa versa. If one computer is compromised through a shared network drive, the Windows password file can be obtained. This is encrypted but this encryption can easily be broken and thus passwords for neighbouring computers can be obtained. The difficulty involved in remembering several passwords causes people tend to reuse the same password on different computers and accounts, and this can result in a “password nesting” attack to all computers that are connected to the compromised one.

5.3.3 Policies and requirements

5.3.3.1 Knowledge and awareness Users should be educated about security, network knowledge and the importance of awareness as much as possible. This should include the importance of passwords and smart cards for authentication purposes, consequences of attacks, various threats and security measures (anti-virus software, firewalls, IDS, resource sharing). This applies to home networks as well as in a general context. “Code Red” and its variants have shown just how badly compromised machines can affect the network, and hand control of a machine to an unauthorised person for nefarious purposes. Improved security at the users’ premises - on the home networks in particular, decreases the chances for further attacks once a host on a particular home network is compromised. While this may not be an obvious consequence if a Windows operating systems based host is attacked, it will definitively be the case if the host in question is running a multi-user operating system, based on shell accounts, such as UNIX and LINUX. However, one can to some extent assume that users with relatively good computing knowledge and ability to protect their servers will install such operating systems (see section 5.3.5).

5.3.3.2 Firewalls and network architecture Users can establish home networks either by having multiple accounts for each service available or by using their own NAT software (or hardware). In the first scenario each device on the home network is connected to a hub, which in turn is connected to the CPE (= xDSL modem) or RG (in the case that the CPE is fully integrated in the RG). In this case, it would be necessary to install firewall software on each computer if firewalls were to be employed. This will require maintenance and configuration of each firewall. In the second scenario one single computer acts as a proxy server running NAT and firewall services. Every other device is connected to a hub, which in turn is connected to the proxy, which finally is connected to the CPE/RG. With this configuration only one firewall needs to be maintained and configured, but anti-virus should still be run on each of the other computers. With this architecture NAT will contribute to hiding the IP addresses of the other computers, thus all connections to the outside will seem as if they were made from the proxy. However, with this configuration, the proxy server will remain visible from the Internet.

______35 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

It is also possible to implement NAT and Firewall services in the CPE or RG, and it is recommended that TORRENT develops and tests the integration of these services in the RG. It is possible to configure these services so that none of the local computers are directly externally visible from the Internet. The use of NAT and firewall services in the RG will mask the home network users’ IP addresses from others on the Internet making it more difficult for a hacker to target and attack machines that are located within the home network. Certain ports can be blocked and redirected to limit the services that outside users can access. The user can open specific ports to ensure that games and other Internet applications will run properly.

5.3.4 Anti-virus software Anti-virus software and a firewall should be considered as a minimum security measure independent of configuration, but preferably secondary security measures such as Intrusion Detection Systems should be employed too.

5.3.5 Suggested home network policy elements Users should strive to improve knowledge and awareness and, wherever possible, the operators and service providers should educate customers about security issues. Most important is to be aware that there is need for security, what to protect, how to protect it according to the chosen home network architecture and be aware of the consequences of an intrusion or attack from the outside. This includes looking out for any irregular behaviour and obtaining knowledge about signs of attack. Part of the awareness aspect is to keep software up to date. Security holes in well-known applications are discovered on a regular basis, and patches and service packs are released accordingly. The users should pay attention and apply these as they are released. This is especially important for anti-virus software as new viruses are discovered almost every day. Shared resources and running services should be protected with “good” passwords. Names, names of places, pets and similar are typical “bad” passwords. There are several guidelines for composing a “good” password. Passwords should be changed on a regular basis. Users should analyse their needs and shut down services running on the home network that are not in use. The more running services - the more ways to break into the network. In a Windows operating system environment, one should pay particular attention to the configuration of the file and printer sharing service. Unless needed, this service should be unbound from the TCP/IP protocol and rather be bound to use a non-routable protocol such as NetBEUI or IPX/SPX. This prevents the entire Internet from being able to access the shared resources directly. Security measures should be applied. A suggested minimum is some sort of firewall protection. The amount of protection a firewall can offer is highly dependent on the computing and networking knowledge of the users. A dedicated firewall/proxy or packet filter will probably offer better protection in accordance with the specific needs of a particular home network than a predefined set of rules in “out of the box”-software. Still this must be considered as solutions for advanced users. Additional security measures should be applied where available. Anti-virus software and intrusion detection software are examples of this.

______36 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

If anti-virus software, IDS or firewall reports possible attempts on break-ins, users should have a set of actions to perform. This will to a great extent be different for each host on the home network and for individual needs, but in general: Whenever one becomes aware of a compromise or intrusion on a host on the network, it should be disconnected from the network to prevent further damage while it is inspected. Anti-virus software should be applied and firewall logs inspected (if available). If a virus, such as a Trojan, is detected, all passwords that have been typed in or have been in use on that particular computer should be changed. Preferably also any other passwords that have been transmitted in clear-text on the home network should be changed. If abnormal behaviour or other signs of intrusion occurs in an environment with a device using a closed service, the particular host that is connected to the closed service should be shut down for a period of time and event-specific measures should be applied.

______37 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

6. The Residential Gateway

This section describes the potential impact of TORRENT Residential gateway (RG) as well as its main features and functions without referencing internal mechanisms. Existing or emerging RGs differ according to the network and service capabilities they offer (eg. [1-4]). Even though several commercial systems exist and others are being specified (or prototyped), there is no widespread deployment of those systems, while provisioning of broadband services to the home is still in its infancy. This is due to several reasons. Among others, existing products are not compliant to a common specification, currently offered systems are owned by service providers and/or targeted for a particular access network or in-home service, while potential customers are not aware of the capabilities and elegance of these systems. The TORRENT RG (or simply RG) is a network device that enables a wide range of broadband services and applications to the home. This is achieved by allowing multiple users at home to simultaneously access: • public Internet (by sharing a single broadband connection and ISP account) and • in-home network resources (eg., by sharing a wired or wireless LAN). The RG supports standard communication interfaces that ease the connection to both WAN and in-home network infrastructures; ie. the RG distributes a high-speed broadband connection to a series of home networking technologies, enabling home client terminals (eg. PCs, notebook computers, palmtops) to gain access to video, voice, and data services. At the same time, expensive peripheral devices (eg. printers), audio/visual equipment, and other home appliances can be accessed through home networking. The RG is a fully scalable system in terms of functionality. Downloadable software modules will provide potential services/applications, while the system allows for future extensions/upgrades in a straightforward fashion. Routing, security, VPN and QoS support, and application sharing are core functions of the RG. Service providers are mainly concentrating on the widespread deployment of new services to the home. Broadband access networks technology when coupled with RGs will enable the delivery of new broadband services to the home incorporating video and voice integration with data over a single high-speed connection. This will result in more revenue for service providers. Furthermore, service providers impose requirements for remote updating and upgrading to both ease the offering of new services and eliminate the cost of local maintenance. In addition, they are looking for an easier and more cost-effective method to manage their network infrastructure. Equipment manufacturers mainly impose requirements for an agreed RG specification. This will eventually lead to a high penetration and short time to market. In this way, the RG will become a commodity, while equipment manufactures will reduce their development costs. On the contrary, nowadays, the lack of a standard results in proprietary products, and often in a non-affordable cost.

6.1 High-level System Requirements This sub-section rationalises from a high-level point of view the impact of the RG on the Internet access and the home networking.

6.1.1 Impact on Internet access With respect to Internet access, the RG should support the following: ______38 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

• interworking with various access networks • easy and secure sharing of a high-speed broadband connection • Internet access for various home clients irrespective of their operating system • existing and future Internet applications • scalability in functionality • reliability in operation • easy configuration • easy upgrading • easy network management • compliance with existing standards and RFCs.

6.1.2 Impact on home networking With respect to Internet access, the RG should support the following: • typical home networking technologies (both wired and wireless) • interconnection of home networking devices • sharing of network resources • server functionality • provisioning of an application execution environment for existing and future applications (eg. data, entertainment, control).

6.1.3 Functional requirements The main functional requirements for the RG are described below. A common basis for the following is the support of a high-performance TCP/IP stack (IPv4 is mandatory; IPv6 is optional).

6.1.3.1 Routing Broadband Internet connection sharing: The RG should provide a networking function to allow in-home computers connected to the RG to access Internet over a single broadband connection, even though only one official IP address exists (allocated to the RG). Such a function is similar to NAT (Network Address Translation) found in many commercial firewalls and industrial routers. IP routing and forwarding: The RG should provide IP routing and forwarding capabilities between the WAN and in-home network interfaces. Dynamic allocation of IP addresses: Each home computer will be assigned a private IP address (as per v4 networking). The allocation of addresses for clients on the home networks will be performed dynamically by a DHCP server running on the RG. In addition, a DHCP client should be present to fetch an IP address from nearly every ISP. IPv6 support (optional): When IPv6 networking is enabled, each networking device at home will obtain an IPv6 address, thus, no need for private addresses exists. Further, addresses are assigned to clients by either stateless autoconfiguration or DHCPv6.

______39 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

6.1.3.2 Security Due to the always-on feature of a broadband connection between the home premises and the Internet, home computers are prone to outside attacks. Thus, security is an important aspect that must be considered. The RG must secure home computers from intruders. Security options must include firewall, virus protection and/or any other intruder detection system. Virus protection is essential. Virus protection software is installed on the PC and provides protection by checking files and programs for viruses. For IPv4 networking, NAT is applied as well that allows for the export of one IP address to the global Internet, that of RG. However, use of NAT will introduce challenges for Service Providers, as there are difficulties involved in getting services (eg. SIP services) through residential NATs.

6.1.3.3 Layer 2 support The RG should provide for Layer 2 support, in particular PPP over WAN links and ARP.

6.1.3.4 VPN support VPN support should be provided mainly by the IPSec, PPTP, and L2TP protocols.

6.1.3.5 QoS support QoS support (especially for real-time traffic) should be provided by packet scheduling on the RG. In addition, prioritisation of Ethernet traffic may be provided.

6.1.3.6 Network management Network management and statistics should be provided via the SNMP protocol. A web-based front-end interface to SNMP should also be provided to facilitate local or remote interaction with the RG system. In addition, the RG should provide means for both local (eg. serial) and remote configuration (eg. telnet).

6.1.3.7 Server functionality The RG should provide the capabilities of a file, print, and application server.

6.1.3.8 Application execution environment The RG should provide an application execution environment in accordance with [5].

6.1.3.9 External network interface requirements The following network interface options are identified for the RG:

Access A broadband connection to the home can be provided by an ADSL modem, cable modem, wireless, Long- Reach Ethernet, or satellite interface. It is also noted, that an ISDN 128 kbits/s dial up connection is also still an acceptable solution. It is common for existing access network termination devices to be equipped with an Ethernet (or USB) port.

______40 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

In-home In-home communications are realised with the following technologies: Ethernet, IEEE 802.11b (wireless LAN), HIPERLAN/2 (wireless access system), Bluetooth (short range wireless system), and HomePNA (utilises existing phone lines).

Other communication ports Standard serial/USB ports.

6.1.4 Basic interface configuration Since several network interface options exist, a basic configuration for the RG should be as follows: • One general-purpose 10BaseT Ethernet interface for connection to an external ADSL modem with an Ethernet port. • One 10/100BaseT interface for connection to the home LAN. • One wireless interface of IEEE 802.11b or HIPERLAN/2 type. • One USB port, eg., for connection to an external ADSL modem with a USB port. • One serial port for “command-line” configuration mode of the RG.

6.1.5 Performance issues The RG must not introduce a bottleneck.

6.1.6 System requirements

6.1.6.1 Hardware key features The RG should be implemented in the form of an industrial type networking box. The core of that box should be a standard single board computer (SBC) with off-the-shelf components. That is, devices found in a standard PC, such as monitor and keyboard (mouse optional). As seen from outside, the interfaces described in section 6.1.3.9 will be solely available.

6.1.6.2 Embedded software key features Embedded software (firmware) should be hosted on the hardware platform. The core software platform includes the operating system, Internet protocol stack, protocol daemons, and drivers for the supported devices. A distributed execution environment for applications couples the core software platform.

6.2 Examples of possible RG types It is general considered that current ADSL solutions are not assumed to be cost effective in the long term because the implementations include an excessive number of physical boxes (ie. separate ADSL modem, ISDN NT1 and/or separate line filters), which increase equipment and installation costs. An implementation ______41 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers with multiple boxes (maybe requiring multiple power outlets and a web of interconnecting cabling) is also less competitive because it is esthetically offensive to the average residential subscriber.

Many different implementations are possible, 6 selected examples are described following:

RG1: Services, service requirements Access solution User interface Internet access, home office, video on ADSL: - IP on Ethernet demand: - 512/128 kbit/s - Real time: non critical - 1024/256 kbit/s (optional) - Security: non critical - 2048/448 kbit/s (optional) - 6144/448 kbit/s (optional) Voice and multimedia telephony: - ISDN Basic Rate Interface - ISDN S-bus - Real time: critical - Security: important Alarms remote control, credit card - X.25 in ISDN Basic Rate - ISDN S-bus terminals: Interface D-channel - Real time: non critical - Security: critical Support of legacy analogue user - ISDN Terminal Adapter - Analogue telephone terminals (obsolete service, non- interface profitable in the long-term, but currently required)

Physical External Network configuration Functional implementation copper imple- pair * Note: with an IP router, more functionality (e.g. QoS, firewall from PSTN/ISDN X.25 Internet RG1 mentation network network applications etc.) can be provided network

Supervision ISDN exchange with The Internet and alarms RG1 V5.2 access *Can also be LMDS ISDN ADSL 230 VAC NT1 modem U LINE POWER WDM/SDH/PDH transmission network = DC 230V Analog V5.2 termination Internet access router Ethernet telephone Service access switch/IP ISDN LT ADSL modem ISDN Power set A S2 S1 E2 E1 TA router* supply Line filter Line filter point ISDN Internet LLUB cross connect user A S2 S1 user terminal E2 E1 External 230V AC terminal ISDN S BUS ETHERNET 2.5 km copper pair in the access network* Analog ISDN S bus Ethernet in copper pair DISTRIBUTION DISTRIBUTION telephone in subscriber subscriber from network RG1 Subscriber premises interface premises premises

Explanation: An ordinary ISDN + ADSL access solution physically integrated into a Residential Gateway.

RG Terminal type 1 is assumed to be the currently most cost effective solution. The implementation will have moderate cost and moderate revenue combined from several sources.

______42 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

RG2: Services, service requirements Access solution User interface Internet access, home office, video on G.704 structure: - IP on Ethernet demand: - 512/128 kbit/s - Real time: non critical - 1024/256 kbit/s (optional) - Security: non critical - 2048/448 kbit/s (optional) - 6144/448 kbit/s (optional) - 24/2 Mbit/s (optional) Voice and multimedia telephony: - ISDN Basic Rate Interface - ISDN S-bus (2B + D in G.704) - Real time: critical - G.703 - Security: important - ISDN Primary Rate Interface (partial) (6B+D in G.704) Alarms remote control, credit card - X.25 in ISDN Basic Rate - ISDN S-bus terminals: Interface D-channel - Real time: non critical - Security: critical Support of legacy analogue user ISDN Terminal Adapter - Analogue telephone terminals (obsolete service, non- interface profitable in the long-term, but currently required) CATV: Sub carrier multiplexing Coaxial cable - Real time: non critical - Security: non critical

Functional implementation External Network configuration Physical optical fiber * Note: with an IP router, more functionality imple- from RG2 (e.g. QoS, firewall etc.) can be provided network PSTN/ISDN X.25 Internet CATV Headend mentation network network applications 1550 nm SCM transmitter G.704 mux Supervision and alarms RG2 230 VAC ISDN exchange with ISDN BA ISDN PRA F1 decode The Internet Fiber Amplifier (EDFA) Opt. POWER V5.2 access NT1 NT 1550 nm 1310 nm receiver transmitter WDM/SDH/PDH CATV optical fiber = DC 230V Fn decode Analog transmission network transmission network telephone Ethernet Service Power Coax V5.2 termination ISDN LT Internet access router 1550 nm receiver ISDN switch/IP set A S S/T E FE supply G.704 mux, SCM coder/decoder SCM filter TA router* CATV transm. CATV output access Opt. filter ISDN 1310 nm receiver 1550 nm transmitter ISDN S BUS Internet point BA user Line filter A S S/T E FE Coax Optical fiber FAST user terminal 230V AC terminal ISDN PRA ETHERNET ETHERNET LLUB fiber cross connect Analog ISDN BA ISDN IP on Ethernet CATV output from network telephone S bus partial /Fast Ethernet interface PRA Optical fiber in the access network RG2 Subscriber premises

Explanation: A single optical fibre access solution physically integrated into a Residential Gateway. Bidirectional optical transmission is used, with a single 1310 nm wavelength upstream and a single 1550 nm wavelength downstream. Sub carrier multiplexing in the electrical domain is used to separate the downstream low frequency band supporting a G.704 type frame for ISDN and Internet services and the higher frequency bands ______43 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers supporting traditional (digital) CATV. An additional G.703 port supporting partial ISDN PRA for high quality real time services (eg. videoconferencing) is included.

RG Terminal type 2 is assumed to be the potentially most cost effective solution within a near future. The implementation will have high cost, but also high revenue combined from several sources.

RG3: Services, service requirements Access solution User interface Internet access, home office, video on “Ethernet on fibre”, with IP - IP on Ethernet / Fast demand: mobility Ethernet / WLAN - Real time: non critical - Security: non critical Voice and multimedia telephony, with “Ethernet on fibre”, with IP - IP on Ethernet / Fast moderate quality: priority, IP mobility and Ethernet / WLAN - Real time: important Internet applications in user terminals: - Security: non critical SIP / H.323 / MEGACO + Q.931 on SIGTRAN / MEGACO

Physical External Network configuration Functional implementation fiber imple- from mentation network Internet applications, RG3 including IP telephony (voice and multimedia)

Radio antenna RG3 The Internet Supervision 230 VAC and alarms POWER Synchronous frame OPT WDM/SDH/PDH transmission network Radio termination transceiver 230V = DC Internet router Service access E FE Synchronous frame termination Optical transceiver IP router with support of Optical point IP priority, IP mobility, Power wireless LAN transceiver supply ETHERNET Internet user terminal, LLUB fiber cross connect DISTRIBUTION e.g. IP telephoneset optical fiber in the access network E FE RG3 Subscriber premises External 230V AC Ethernet and Fast Ethernet optical fiber in subscriber premises from network

Explanation: A single optical fibre 10 Mbit/s symmetrical access solution with IP based interfaces physically integrated into a Residential Gateway. Wireless LAN functionality is included to support local mobility. IP mobility is included to support roaming between accesses. IP priority is supported to enable moderate quality real time applications.

RG Terminal type 3 is assumed to be a potentially competitive alternative to RG Terminal type 2 for some market segments within a near future. The implementation will have high cost, but also high revenue combined from several sources.

______44 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

RG4: Services, service requirements Access solution User interface Internet access, home office, video on ADSL: - IP on Ethernet demand: - 512/128 kbit/s - Real time: non critical - 1024/256 kbit/s (optional) - Security: non critical - 2048/448 kbit/s (optional) - 6144/448 kbit/s (optional) Support of legacy analogue user VoIP media gateway, with - Analogue telephone terminals (obsolete service, non- MEGACO support interface profitable in the long-term, but currently required), with low quality, low security VoIP service

Network configuration Functional implementation Physical implementation External copper Internet applications, pair including VoIP RG4 from network The Internet Supervision *Can also be LMDS and alarms ADSL WDM/SDH/PDH transmission network modem RG4 230 VAC Internet access router = DC U LINE POWER ADSL modem Service access 230V VoIP MGW Ethernet Line filter point /MEGACO switch Power supply A E2 E1 LLUB cross connect Line filter 2.5 km copper pair in theaccess network* Analog A E2 E1 External telephone Internet RG4 Subscriber premises 230V AC user terminal Analog Ethernet in copper pair set ETHERNET DISTRIBUTION telephone subscriber from network interface premises

Explanation: A stripped ADSL access solution with VoIP termination physically integrated into a Residential Gateway.

RG Terminal type 4 is not assumed to be cost effective. The implementation will have moderate cost, but low revenue because of insufficient quality for the voice Telephony service.

______45 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

RG5: Services, service requirements Access solution User interface Internet access, home office, video on ADSL: - IP on Ethernet demand: - 512/128 kbit/s - Real time: non critical - 1024/256 kbit/s (optional) - Security: non critical - 2048/448 kbit/s (optional) - 6144/448 kbit/s (optional) Support of legacy analogue user ADSL splitter - Analogue telephone terminals (obsolete service, non- interface profitable in the long-term, but currently required)

Network configuration Functional implementation Physical implementation External copper pair PSTN/ISDN Internet RG5 from network applications

network PSTN/ISDN exchange The Internet Supervision with V5.2 access *Can also be LMDS and alarms ADSL WDM/SDH/PDH transmission network modem RG5 230 VAC = DC U LINE POWER V5.2 termination Internet access router POTS LT ADSL modem Service access 230V Ethernet Line filter point switch Power supply LLUB cross connect Line filter A E2 E1 Analog 2.5 km copper pair in the access network* telephone A Internet E2 E1 External 230V AC set RG5 Subscriber premises Analog Ethernetin copperpair ETHERNET user terminal telephone subscriber from network DISTRIBUTION interface premises

Explanation: An ordinary analog line + ADSL access solution physically integrated into a Residential Gateway.

RG Terminal type 5 is not assumed to be cost effective compared to type 1. The implementation will have moderate cost, but lower revenue because of less revenue generating functionality for the voice Telephony service.

______46 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

RG6: Services, service requirements Access solution User interface Internet access, home office, video on ADSL: - IP on Ethernet demand: - 512/128 kbit/s - Real time: non critical - 1024/256 kbit/s (optional) - Security: non critical - 2048/448 kbit/s (optional) - 6144/448 kbit/s (optional)

Network configuration Functional implementation Physical implementation External

copper Internet applications, pair including VoIP RG6 from network The Internet *Can also be LMDS Supervision and alarms WDM/SDH/PDH transmission network ADSL modem RG6 230 VAC Internet access router U LINE POWER ADSL modem Service access = DC Line filter 230V point LLUB cross connect Ethernet switch Power E2 E1 2.5 km copper pair in theaccess network* supply Line filter RG6 Subscriber premises Internet E2 E1 External ETHERNET user terminal 230V AC DISTRIBUTION Ethernet in subscriber copper pair premises from network

Explanation: A stripped ADSL access solution without support for voice telephony physically integrated into a Residential Gateway.

RG Terminal type 6 is not assumed to be cost effective. The implementation will have moderate cost, but low combined revenue because the voice Telephony service will not contribute significantly to the revenue.

______47 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

7. Access Network Protocols, Interfaces and Functionalities

There are several competing technologies in the access network. The market is enormous (every user needs an access network), and the issue is being addressed by operators and manufacturers both from the point of view of exploiting to the full the legacy networks (eg. using ISDN and xDSL) and introducing new technologies (eg. optical and broadband wireless). Each solution has advantages and disadvantages for users and operators, the degree of importance of which depends upon the user’s service requirements, the location, and whether a legacy network already exists. It is a fundamental goal of TORRENT to minimise the disadvantages by enabling alternative paths when services demand more stringent capabilities.

7.1 Fixed Access Networks

7.1.1 ISDN: Integrated Services Digital Network ISDN is a world-wide standardised system of digital phone connections which has been available for over 15 years. Voice and data are carried by bearer channels (B channels) occupying a bandwidth of, generally, 64 kbit/s (the US implementation limits B channels to a capacity of 56 kbit/s). A data channel (D channel) handles signalling at 16 kbit/s or 64 kbit/s, depending upon the service type.

There are two basic types of ISDN service: Basic Rate Interface (BRI) and Primary Rate Interface (PRI). BRI consists of two 64 kbit/s B channels and one 16 kbit/s D channel for a total of 144 kbit/s. This basic service is intended to meet the needs of most individual users.

PRI is intended for users with greater capacity requirements. Typically the channel structure is of 30 B channels plus one 64 kbit/s D channel for a total of 1984 kbit/s (E1). In the US, the structure comprises 23 B channels plus one 64 kbit/s D channel for a total of 1536 kbit/s (T1).It is also possible to support multiple PRI lines with one 64kbit/s D channel using Non-Facility Associated Signalling (NFAS).

H channels provide a way to aggregate B channels. They are implemented as: • H0 = 384 kbit/s (6 B channels) • H10 = 1472 kbit/s (23 B channels) • H11 = 1536 kbit/s (24 B channels) • H12 = 1920 kbit/s (30 B channels) To have a BRI service, users must be within 5.5 kms of the local exchange; beyond that, expensive repeater devices are required, or ISDN service may not be available at all. Users also need special equipment to terminate the line and digitalise the voice.

The advantages of ISDN for the residential user are basically: • speed • the possibility to attach multiple - and different – devices to the same access line • the availability of 2 simultaneous channels • supplementary service features.

______48 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

7.1.1.1 Speed The modem was a big breakthrough in computer communications. It allowed computers to communicate by converting their digital information into an analogue signal to travel through the public phone network. There is an upper limit to the amount of information that an analogue telephone line can hold. Currently, it is about 56 kbit/s (ITU-T Recommendation V.90). ISDN, however, allows multiple digital channels to be operated simultaneously through the same regular phone wiring used for analogue lines. This scheme permits a much higher data transfer rate than analogue lines. BRI ISDN, using a channel aggregation protocol such as BONDING or Multilink-PPP, supports an uncompressed data transfer speed of 128 kbit/s. In addition, call setup time is very fast. For example, a V.34 modem typically takes 30-60 seconds to establish a connection; an ISDN call usually takes less than 2 seconds.

7.1.1.2 Multiple devices Instead of the phone company sending a ring voltage signal to ring the bell in the phone ("In-Band signal"), it sends a digital packet on a separate channel ("Out-of-Band signal"). The Out-of-Band signal does not disturb established connections. The signalling also indicates who is calling, what type of call it is (data/voice), and what number was dialled. Available ISDN phone equipment is then capable of making intelligent decisions on how to direct the call. Previously, it was necessary to have a phone line for each device you wished to use simultaneously. For example, one line each was required for a telephone, Fax, computer, bridge/router, and live video conference system. Transferring a file to someone while talking on the phone or seeing their live picture on a video screen would require several potentially expensive phone lines. It is possible to combine many different digital data sources and have the information routed to the proper destination. Since the line is digital, it is easier to keep the noise and interference out while combining these signals. ISDN technically refers to a specific set of digital services provided through a single, standard interface. Without ISDN, distinct interfaces are required instead.

Interfaces The U interface is a two-wire (single pair) interface from the local exchange. It supports full-duplex data transfer over a single pair of wires, therefore only a single device can be connected to a U interface. This device is called a Network Termination 1 (NT-1). The NT-1 is a relatively simple device that converts the 2-wire U interface into the 4-wire S/T interface. The S/T interface supports multiple devices (up to 7 devices can be placed on the S/T bus) because, while it is still a full-duplex interface, there is now a pair of wires for receiving data, and another for transmit data. Today, many devices have NT-1s built into their design. This has the advantage of making the devices less expensive and easier to install, but often reduces flexibility by preventing additional devices from being connected. Technically, ISDN devices must go through a Network Termination 2 (NT-2) device, which converts the T interface into the S interface (Note: the S and T interfaces are electrically equivalent). Virtually all ISDN devices include an NT-2 in their design. The NT-2 communicates with terminal equipment, and handles the Layer 2 and 3 ISDN protocols. Devices most commonly expect either a U interface connection (these have a built-in NT-1), or an S/T interface connection. Devices that connect to the S/T (or S) interface include ISDN capable telephones and Fax machines, video teleconferencing equipment, bridge/routers, and terminal adapters. All devices that are designed for ISDN are designated Terminal Equipment 1 (TE1). All other communication devices that are not ISDN capable, but ______49 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers have a POTS telephone interface (also called the R interface), including ordinary analogue telephones, Fax machines, and modems, are designated Terminal Equipment 2 (TE2). A Terminal Adapter (TA) connects a TE2 to an ISDN S/T bus.

Layer 1: Physical Layer The ISDN Physical Layer is specified by the ITU I-series and G-series documents. The U interface provided by the telco for BRI is a 2-wire, 160 kbit/s digital connection. Echo cancellation is used to reduce noise, and data encoding schemes (see below) permit this relatively high data rate over ordinary single-pair local loops.

Data Encoding The Basic-Rate-Interface and the Primary-Rate-Interface have different line coding schemes. In Europe 4B3T is used for the BRI and HDB3 for the PRI, in the US 2B1Q is used for the BRI and B8ZS for the PRI. These differences result from the differences in the existing hardware in the telephone plants. The European E1 system uses HDB3 and therefore the PRI has the same transmission rate and the same line codes whereas the DS1 system in the US uses 4B3T. 4B3T stands for 4 binary data, 3 ternary redundancy and is a block code in which 4 binary bits are converted into three ternary digits. This code increases the line efficiency. For example, if this code were to be employed on a 140Mb/s digital system the signalling rate would change from 140Mb/s to 140*3/4= 105Mbauds/s, requiring reduced bandwidth. Blocks of four binary digits give 16 code words whereas the 3 ternary digits give 27. Hence, as only 16 codewords are required to recode the binary codewords there are 11 code words in the ternary system left redundant. HDB3 high-density bipolar order 3 so called because of the increased number of transitions compared with the bipolar signal from which it is derived.. This code is a bipolar signalling technique (ie. relies on the transmission of both positive and negative pulses). It is based on Alternate Mark Inversion (AMI), but extends this by inserting violation codes whenever there is a run of 4 or more 0's. 2B1Q stands for 2 binary data, 1 quaternary (1988 ANSI spec T1.601). A quaternary coding scheme uses four different voltage levels, each representing a group of two bits. The first bit is called sign bit and determines whether the voltage is positive or negative. The second bit is called a magnitude bit and determines whether the voltage is 1V or 3V.

Quaternary Voltage Bits Symbol Level 00 -3 -2.5 01 -1 -0.833 10 +3 +2.5 11 +1 +0.833

This means that the input voltage level can be one of 4 distinct levels (note: 0 Volts is not a valid voltage under this scheme).

B8ZS stands for binary with-8-zeros substitution coding and is based on Alternate Mark Inversion. The difference is that B8ZS substitutes a series of 8 zeros by the following series: the first three are unchanged, violation, correct 1, 0, violation, correct 1. Again these artificially introduced transitions support the synchronisation at the receivers side. ______50 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Figure 7.1: Coding Schemes Overview

The ISDN Data Link Layer is specified by the ITU Q-series documents Q.920 through Q.923. All of the signalling on the D channel is defined in the Q.921 specification.

Frame Format Each U interface frame is 240 bits long. At the prescribed data rate of 160 kbit/s, each frame is therefore 1.5 msec long. Each frame consists of: • Frame overhead - 16 kbit/s • D channel - 16 kbit/s • 2 B channels at 64 kbit/s - 128 kbit/s

Sync 12 * (B1 + B2 + D) Maintenance 18 bits 216 bits 6 bits

• The Sync field consists of 9 quaternaries (2 bits each) in the pattern +3 +3 -3 -3 -3 +3 -3 +3 -3.

• (B1 + B2 + D) is 18 bits of data consisting of 8 bits from the first B channel, 8 bits from the second B channel, and 2 bits of D channel data. • The Maintenance field contains CRC information, block error detection flags, and "embedded operator commands" used for loopback testing without disrupting user data. Data is transmitted in a superframe consisting of 8 240-bit frames for a total of 1920 bits (240 octets). The sync field of the first frame in the superframe is inverted (ie. -3 -3 +3 +3 +3 -3 +3 -3 +3).

LAP-D Link Access Protocol - D channel (LAP-D) is the Layer 2 protocol used. This is almost identical to the X.25 LAP-B protocol. The structure of a LAP-D frame is as follows:

Flag Address Control Information CRC Flag ______51 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Flag (1 octet) - This is always 7E16 (0111 11102)

Address (2 octets) 12345678 SAPI (6 bits) C/R EA0 TEI (7 bits) EA1

SAPI (Service access point identifier), 6-bits (see below) C/R (Command/Response) bit indicates if the frame is a command or a response EA0 (Address Extension) bit indicates whether this is the final octet of the address or not TEI (Terminal Endpoint Identifier) 7-bit device identifier (see below) EA1 (Address Extension) bit, same as EA0 Control (2 octets) - The frame level control field indicates the frame type (Information, Supervisory, or Unnumbered) and sequence numbers (N(r) and N(s)) as required. Information - Layer 3 protocol information and User data CRC (2 octets) - Cyclic Redundancy Check is a low-level test for bit errors on the user data.

Flag (1 octet) - This is always 7E16 (0111 11102)

SAPIs The Service Access Point Identifier (SAPI) is a 6-bit field that identifies the point where Layer 2 provides a service to Layer 3. See the following table:

SAPI Description 0 Call control procedures 1 Packet Mode using Q.931 call procedures 16 Packet Mode communications procedures 32-47 Reserved for national use 63 Management Procedures Others Reserved for Future Use

TEIs Terminal Endpoint Identifiers (TEIs) are unique IDs given to each device (TE) on an ISDN S/T bus. This identifier can be dynamic; the value may be assigned statically when the TE is installed, or dynamically when activated.

TEI Description 0-63 Fixed TEI assignments 64-126 Dynamic TEI assignment (assigned by the switch) 127 Broadcast to all devices

______52 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Layer 3 Call Setup These are the steps that occur when an ISDN call is established. In the following example, there are three points where messages are sent and received; 1) the Caller, 2) the ISDN Switch, and 3) the Receiver. 1. Caller sends a SETUP to the Switch. 2. If the SETUP is OK, the Switch sends a CALL PROCeeding to the Caller, and then a SETUP to the Receiver. 3. The Receiver gets the SETUP. If it is OK, then it rings the phone and sends an ALERTING message to the Switch. 4. The Switch forwards the ALERTING message to the Caller. 5. When the receiver answers the call, is sends a CONNECT message to the Switch 6. The Switch forwards the CONNECT message to the Caller. 7. The Caller sends a CONNECT ACKnowledge message to the Switch 8. The Switch forwards the CONNECT ACK message to the Receiver. 9. Done. The connection is now up.

7.1.2 XDSL Technologies (HDSL, SDSL, IDSL, VDSL, ADSL) To offload Internet traffic from the switches at the edge of the PSTN, it is recommended that operators promote the deployment of xDSL services, particularly to heavier users (eg. home workers, home offices, small offices), so that they are encouraged to move from dial up connectivity (which consumes TDM resources on the Level 4 switches) to dedicated xDSL access which is not carried on the PSTN.

The xDSL service is connected to exchange servers to permit interaction between the exchange services and the PC and video terminals.

Exchange Servers Analogue and Extensions Home terminal drivers

Server Multimedia Extension 64kbit/s Multimedia Phone 64kbit/s with Digital Interface LAN Modem64kbit/s Modem Modem Copper Pair (at least 64kbit/s)

Server PC with Internet (in place of phone data)

xDSL xDSL Router Service Service PC (Internet) TV Internet Video Figure 7.2: Multimedia Service ______53 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Local Access Overview: • The architecture of the servers in the local exchange could be based on a modem bank connected to a 100 BaseT LAN that is connected to two different servers: • Provides load sharing • Redundancy • The exchange modem bank could be a DSP unit that serves multiple lines as the dial-in units do today: • The exchange modem link is similar technology • The primary subscriber terminal could be a single modular unit that contains the modem and all the interconnection points for the other services. • It serves as the basic telephone and can be upgraded to include the display screen • Extension terminals are the same as the primary terminal, but have a different modem for local data services (see Home Network Concept).

There are several types of xDSL technologies: 1. HDSL High speed digital subscriber line was the first version of DSL introduced to resolve the increasing need for high speed connections. It provides a full duplex DS1 over 12000 feet on two pair of telephone wires. HDSL can be installed using one pair supporting half the data rate if the need arise. Neither the full nor half data rate can be considered as the optimum solution for widespread use, due to the following issues: • The need for repeaters in the service loop in order to maintain 1.5 to 2Mbit/s data rate (cost prohibitive) • The use of two twisted pair • The use of voice band frequency HDSL and propriety SDSL systems do not use the available bandwidth in the copper lines very efficiently and cause spectrum compatibility problems with other DSL lines in the same cable bundle. In reality with the Local Loop Unbundling, different operators will use the same copper bundle. As a result, the regulator will not allow anymore the use of non-spectrum friendly DSL technologies in the local loop. To solve the spectrum compatibility problem with the old HDSL/SDSL technologies new more sophisticated Symmetrical DSL technologies are currently being standardised in the different standardisation bodies. ANSI has defined HDSL-2 for fixed T1 rates and ETSI has defined a SDSL standard further on called ETSI-SDSL to avoid all confusion with any proprietary SDSL. 2. SDSL SDSL (Symmetrical Digital Subscriber Loop) is a name used to indicate a wide range of proprietary DSL technologies used by some vendors to offer higher bandwidths on a single copper pair. In most cases, it is derived from an older HDSL technology that allowed the transport of 2Mbit/s over two or three copper pairs. Both HDSL-2 and ETSI-SDSL will finally converge into the ITU G.SHDSL standard with annexes defining the ANSI or ETSI specific differences, mainly limited to different Power Spectral Densities (PSD). ______54 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

The ETSI-SDSL enables symmetric services at speeds up to 2.3 Mb/s over unconditioned copper. The ETSI-SDSL solution is specifically adapted for the delivery of voice and data services to small and medium sized businesses. The Recommendation G.SHDSL describes a transmission method for data transport in telecommunications access networks. G.SHDSL transceivers are designed primarily for duplex operation over mixed gauge two-wire twisted metallic pairs. 3. IDSL Stands for ISDN DSL, it provides 128KB/s data rate of symmetrical bandwidth on a single pair, and it can cover spans of up to 18Kft. IDSL is a redundant of ISDN, the only difference is instead of being terminated on an ISDN switch, it is terminated on a router and passes the traffic to the Internet. 4. VDSL VDSL (Very high speed DSL) is the ultimate solution for broadband services over copper telephone lines. It supports up to 51.84Mbit/sdown stream on a maximum distance of 1000 ft (the data rate decreases to 25.92Mbit/s for a 3000 ft run). The upstream rates are in the range of 1.5 to 16Mbit/s. VDSL is deployed from the cabinet or from the local exchange, referred to as Fibre to the Cabinet (FTTCab) and Fibre to the Exchange (FTTEx) respectively. VDSL is essentially a fibre to the node architecture with an optical network unit sited in the access network. Symmetric and asymmetric modes of operation are possible. The up- and downstream rates depend upon the loop length, the noise environment and the selected downstream-to-upstream ratio. VDSL has to be spectrally compatible with existing xDSL systems like ADSL, ISDN-BRA (Basic Rate Access), ISDN-PRA (Primary Rate Access) and HDSL (High-speed DSL). 5. ADSL On the asymmetric DSL technologies special emphasis should be put on ADSL (Asymmetric DSL). ADSL is considered to be the optimum choice for personal broadband connection due to capability of delivering full motion video and coexistence with voice services: ADSL combines the benefits of the DMT (Discrete Multi-Tone) and ATM (Asynchronous Transfer Mode) technologies, resulting in:

• 6Mbit/s over 12K feet and 8Mbit/s over 6.5K feet to the subscriber. The upstream data rates currently being offered are within 64 to 640kbit/s.

• Full bandwidth flexibility: upstream and downstream bit rates can be chosen freely and continuously up to the maximum physical limits. At initialisation, the system automatically calculates the maximum possible bit rate, with a predetermined margin. The service management system can then set the bit rate to the level, determined by the customer service profile, thus maximising noise margin and/or minimising transmit power.

• Full service flexibility: a random mix of services with various bit rates and various traffic requirements (guaranteed bandwidth, bursty services) can be supported, within the available bit rate limits. The DSLAM interfaces to the backbone network (data backbone network, ATM network), and provides a transparent connection of the subscriber line to the PSTN network. POTS splitters are used to mix analogue Telephony services and digital ADSL services, thereby allowing the traditional Telephony services to coexist with new high-speed services on the same twisted pair. The DSLAM is a rack containing the ADSL line termination boards, POTS splitters and the interface (usually STM-1) with the backbone network.

ADSL signals are added alongside existing Plain Old Telephone Service (POTS) signals on the wires to/from the local exchange. Although POTS and ADSL occupy distinct channels, they might influence one ______55 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

other. In devices such as phones, modems and fax machines ADSL signals can cause audible noise, possibly irritating the voice service. Telephone devices on the other hand can interfere with ADSL signals, causing degradation in data throughput. To avoid this mutual interference, an electronic central splitter or distributed filters need to be installed. As the names imply: a splitter splits or combines signals while filters prevent signals from entering or escaping from the device.

At the user side, an ADSL modem is used at the ADSL output of the splitter. The modem consist at the input of a high pass filter which cuts off all the low frequencies, a receiver R for the high bit rate downstream signal and a transceiver T for the upstream signal. The output interface of the ADSL modem is an Ethernet or an ATM interface and it used for the interconnection with the home network equipment. At the local exchange is also an ADSL modem and a splitter. All modems in the local exchange are concentrated onto a Digital Subscriber Line Access Multiplexer (DSLAM), which multiplexes and routes all the ATM traffic to the core network. Local Exc hange Home User

TçëÝöù íï POTS / ISDN Autocommutatore PSTN / ISDN PSTN / ISDN

Splitter Splitter

ãéá nterface ÂÂ M Modem i õðç - ult PC ipl ñå Modem - ex óßåò er ATU-R PC DSLAM

Figure 7.3: ADSL over Copper Pair

Both splitters consist of passive components, so that in case of power loss the Telephony service will continue to work properly.

ADSL is emerging as an access technology that can be used by operators to provide the full range of services: e-mail, web browsing, FTP, VoIP, networked games, video streaming, video conference, VoD, albeit not all1 with the quality that could be achieved from a dedicated network..

1 The capacity is not sufficient to transmit more than a few highly compressed TV channels, but is acceptable for a VoD entertainment service ______56 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

CLIENT

POTS ISDN

PSTN ISDN

VOICE CLIENT DSLAM GATEWAY ATM BB-RAS ETHERNET SWITCH ADSL MODEMS CLIENT MCU & GATEKEEPER

WEB SERVER STREAMING FTP SERVER Serv er Serv er SERVER MAIL SERVER Serv er VoD SERVER Figure 7.4: All services provided over an ADSL transport function

Figure 7.4 shows the architecture of the ADSL system that will be used in the TORRENT testbed. The client’s PC is connected to an ADSL modem via Ethernet, USB or an ATM interface. The modem is connected to a DSLAM via the normal twisted copper pair that was used for the Telephony service (the phone bandwidth is still included in an ADSL system). The DSLAM is connected to the PSTN via a twisted copper pair and also to the ATM network via an STM-1 interface. The next equipment (BB-RAS) is a Broadband Router that concentrates all the ATM traffic and routes it to the IP network. In the TORRENT testbed, an Ethernet Switch will be used to emulate the network of a content service provider. The specific scenarios for video conferencing and VoIP, video streaming and video on demand are shown below: H.320 CLIENT

POTS ISDN

PSTN ISDN

H.323 VOICE CLIENT DSLAM PC TO PHONE GATEWAY ATM BB-RAS ETHERNET SWITCH ADSL H.323 MODEMS MCU & GATEKEEPER CLIENT PC TO PC

Figure 7.4.1: Video Conferencing and VoIP provided over an ADSL transport function ______57 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

The video conferencing service (eg. Windows Netmeeting) allows end users to make a conference call using video and sound, with the use of H.323 signalling, and also to access the same application at the same time with use of the ITU-T Recommendation T.120. The H.323 MCU (Multipoint Control Unit) enables the connectivity to more than 2 users. The MCU is responsible for the interconnection, and for the decision at which time each user must reserve the communication channel. Also with the use of a H.323/H320 gateway it will be possible a communication of an ISDN user, which supports a H.320 end software and hardware equipment and a user that support a H.323 terminal.

CLIENT MPEG1 S/W DECODER MPEG2 H/W DECODER DSLAM

ATM BB-RAS ETHERNET SWITCH ADSL MODEMS CLIENT

WEB SERVER Server Server CONTENT INDEX VIDEO CONTENT SERVER

Figure 7.4.2: Video on Demand Service provided over an ADSL transport function

Video on demand is an application capable to transmit video content over an IP network. Video on demand consists of two parts: user software and hardware requirements and network requirements. The network side equipment consists of a video content server where all movies are kept, and a web server as an application interface for the users. The user side equipment depends upon the required video quality (MPEG-1, MPEG-2, MPEG-4).

DSLAM

TV ATM BB-RAS ETHERNET SWITCH ADSL MODEM

STREAMING SERVER

CLIENT with MPEG Ser v er DECODER MPEG H/W ENCODER

CAMERA

Figure 7.4.3: Video Streaming provided over an ADSL transport function ______58 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Video streaming is an application possible to transmit MPEG 1 and MPEG 2 streams to multiply users over an IP network with multicast capabilities. The application consists of two parts: one on the user side (decoder) and one on the network side (encoder). The encoder transmits the source, which may be a file or a real-time encoding. Only one stream is transmitted at a time, and each decoder receives it, decodes it, and presents it on the TV.

7.1.3 Powerline Communications (PLC)

Low-voltage grid

Figure 7.5: Overview of the PLC System

All three levels of the electicity distribution network (ie. high2-, medium3-, and low4 voltage) can be used for voice- and data transfer. Electricity distribution companies have been using the transfer frequency technique for carrying data over the high-voltage cables since the 1920s, and over medium-and low voltage cables since the 1930s. Experience with pilot installations has shown that - given a maximum absorption of 20 dB - transfer rates of 2Mbits/s over distances of up to 450 m are possible in a frequency range below 1MHz. If the cables are underground, then there is no interference with other electrical objects. Apart from direct access to end users, this technique can also be used to connect to mobile base stations (which also use a 2Mbit/s interface), thereby making the technique particularly attractive for electricity distribution companies that also have a mobile operating license. The use of the PLC technique offers cost advantages compared to installing new cables; PLC systems are quicker to install and cheaper. The following interfaces are supported: • G.703/704 (2Mbit/s)

2 High = 110’000 – 380’000 volts 3 Medium = 10’000 – 30’000 volts 4 Low = up to 400 volts ______59 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

• n x 64 kbit/s • X.21 • V.35 • V.36 These can be used for voice and data traffic.

Powerline Communication (PLC) offers: • Internet service: bridging the "last mile" to the customer without installing any additional access line. The service package includes broadband (2Mbit/s) data, IP telephony and the usual ISP services, such as electronic commerce, security services, monitoring functions • Ease of installation: use of any power socket for a data connection – no structural changes are required to connect a single terminal, or build a home LAN • Security services, monitoring functions, emergency services • Always on • Standardised interfaces, based on ISDN • Energy & Consumption Management: cost optimisation for reading electricity, gas and water meters • Technical Services: facility management and home automation (including remote control). Direct integration of circuits to PCs, household machines, etc. is planned

Figure 7.6: Overview of the PLC Environment

7.1.3.1 System Architecture The PLC system is essentially a bus structure. Groups of customers share a frequency band, which is further subdivided into TDMA slots.

______60 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Channel absorption increases with the frequency and cable length. Customers that are close to a node can usually transfer more successfully in much higher frequencies than customers further away from a node. The best channel usage can be obtained when the available frequency band is divided into several channels. Every frequency channel is given to the specific group of customers that have similar cables and therefore similar transfer conditions. It is possible to change the frequency band to which a customer belongs if., for example, he is experiencing interference from a narrowband user. The node functions as an intelligent network manager

and distributes the frequency bands (see Figure 7.7).

Increasing cable length

Channel absorption in dB in absorption Channel

DOWNSTREAM DOWNSTREAM DOWNSTREAM DOWNSTREAM

UPSTREAM UPSTREAM UPSTREAM UPSTREAM

1 Frequency in MHz 20

Figure 7.7: The principle of the division of channels with FDMA. (In order to obtain an optimal usage of the channels, the available bandwidth is divided into several frequency slots)

Since every FDMA-Frequency channel is used by several customers, a multiple access process must be implemented in every channel. Every channel has to have a bandwidth of several Mhz to allow a data rate of about 2Mbits per second within one FDMA frequency channel. Separate channels are used for the downlink and uplink, since the load of interference is dependent on the location and therefore the transfer quality can be different in both directions. Downlink: The downlink requires the highest capacity. It would therefore be an advantage to allocate, if necessary, a complete FDMA frequency channel to every user. Since the node has total control of the preparation of the data to be transferred, it is logical to use a TDMA process within a FDMA part-channel for the multiple access. Figure 7.8 shows the formatting of the data through the node.

______61 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

User 1 User 3

Master … 121 3 … User 2

Data Address

Figure 7.8: The transmission from the master node to the users with TDMA. (In the case of a downlink, it is possible to obtain a very flexible distribution of the systems resources to the users through TDMA)

The bitstreams for the individual users are divided up into data bursts and are sent after each other. The active users are not allocated fixed time slots, but rather the node gives every datagram the address of the recipient of the data burst. Therefore the user only has to check the address of the packet received and can then extract the data addressed to him. The node can send different amounts of datagrams to all users, and this is where the flexibility is needed in the data rate per user. The burstlike transfer doesn't contradict the intention of also using the system for speech transfer. A speech data stream in burst form can be achieved by using appropriate intermediate storage and through the usage and identification of talk breaks. To avoid delays in transferring, speech data can be handled with a higher priority than normal data packets. To avoid blockages of data connection through telephone connections, the system operator can limit the maximum amount of simultaneous telephone calls. Uplink: The uplink path from users to the node differs by the expected average bit rate. On the uplink, especially for Internet access, user normally sends much less data than he receives on the downlink. A burstlike transfer on the uplink is much more difficult to implement than on the downlink. This is because there is no common synchronisation channel that coordinates the formatting of the data packets between the users. There are two possibilities for access to the channel for the users. One variation is the random access of the channel. Collisions can occur which may make it necessary to repeat lost datagrams. This makes random access inappropriate for speech transfer. The node must additionally give fixed assignments to the random access for speech connections. The other variation is to use demand assignment. Using this access principle, every user asks the node for system resources when necessary, and collisions don’t occur. As in the downlink, telephone connections will be handled with higher priority than normal data connections. The advantage of demand assgnment is a higher data throughput, as long as high performance processors can be implemented, that allows a fast demanding and distribution of system resources. An alternative to TDMA is CDMA (code division multiple access). Every user has an individual code to spectrally spread his data. This way the system can always separate the users’ signals. By using different lengths in the spread codes one can achieve the needed flexibility in the data rate. It is not recommended to use the CDMA for the downlink, since high-performance CDMA receivers are very complex and the user’s modems should be kept as simple as possible due to cost reasons.

______62 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

7.2 Optical Fibre (glass or plastic)

Figure 7.9: Illustration showing the application areas of Passive Optical Networks

A Passive Optical Network (PON) is an optical access architecture that facilitates broadband communications in the last mile of the communications infrastructure between the service-provider central office, head-end or point of presence, and business or residential customer locations. These applications are called respectively Fibre-to-the-Curb (FTTC), Fibre-to-the-Building (FTTB) and Fibre-to-the-Home (FTTH). Passive optical networks aim to break the last mile bottleneck by targeting for speeds up to STM-1, that other access technologies do not adequately address. The two primary types of passive optical network technologies are ATM PONs and Ethernet PONs. Network being passive means that the network does not have active elements in the network path, such as lasers, regenerators and amplifiers. Instead passive fibre optic couplers and splitters are used in order to optically route traffic. By reducing or eliminating the number of active components in the network (such as lasers, regenerators and amplifiers), PONs are cutting costs and maintenance, and improving network performance. The architecture is also fibre-efficient.

______63 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Figure 7.10: Placement of the ONU in different topologies

PONs consist of an Optical Line Terminal (OLT) located at the central office or cable head-end, and multiple remote Optical Network Units (ONUs) that deliver broadband voice, data and video services to the end users. The ONUs are geographically distributed in buildings, on curbs, on utility poles or on the sides of homes. One of the primary functions of the ONU is to receive traffic in an optical format and convert it to the customer’s format. A high bandwidth optical signal is sent on a single optical fibre line, and then optically split to several ONUs. An ONU receives and transmits an independent optical frequency, and provides end users with dynamically allocated bandwidth for voice, data and video services. The sending and receiving of data can be transmitted over the same fibre strand, or over two independent strands. The active elements such as the optical line terminal (OLT) and multiple optical network units (ONU) are located at the endpoints of the PON. Upstream traffic from different Optical Network Units is time division multiplexed onto the PON. PONs can be asymmetric in the sense that the downstream bandwidth is higher than the upstream bandwidth. PON is by nature a downstream point-to-multipoint broadcast media, and broadcasts signals from the OLT at the Central Office downstream to every ONU. This makes PONs an appealing solution for telephone companies who plan to introduce cable television.

______64 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Figure 7.11: Illustration showing the association between the OLT and several ONUs.

The PON technology enables carriers to serve customers quickly without adding miles of new fibre to each individual customer, is particularly attractive as a broadband solution for small and medium-sized enterprises, as well as residential subscribers. PONs are also well suited for metro edge applications. Since fibre optics provide a virtually unlimited amount of bandwidth, the same PON infrastructure easily supports higher rate services needed by larger, “high capacity” customers. As a result, carriers can economically provision “T1/E1 type replacement” services to small and mid-sized customers, while at the same time sell high-bandwidth services for applications such as Storage Area Networking - all on the same fibre infrastructure.

Rather than competing with DSL, cable modems and local multipoint distribution system (LMDS), PONs complement these technologies by serving as a feeder to extend fibre from the local exchange to the neighbourhood or curb, where copper, coaxial or wireless systems provide the last-mile connections to the subscribers. To provide FTTC or FTTB, PONs can be used between the local exchange and the Remote Terminal. To provide FTTB or FTTH, PONs can be used between the Remote Terminal and the Customer Terminal Equipment.

Traffic from the ONUs is aggregated back to the OLT using network topologies such as tree, bus or fault- tolerant rings. Whether the network architecture is a protected ring topology or non-protected tree or bus topology, the system offers high bandwidth and ease of provisioning and deployment.

Figure 7.11.1: PON (ring architecture) ______65 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Figure 7.11.2: PON (tree architecture)

Figure 7.11.3: PON (bus architecture)

Unlike an active network, which requires installation of all nodes up front (because each node is a regenerator), PONs can be deployed incrementally on a pay-as-you-go model. This difference means that PONs require less up front investment; a carrier can deploy initially only the fibre infrastructure and the minimum set of equipment required to meet the service requirements. ONUs can be added incrementally as the demand for service grows. Another positive aspect of PONs is that the passive optical components are long-lived, therefore they reduce maintenance in the outside plant. A PON solution also requires a minimum amount of fibre. Using a single fibre strand for multiple ONUs provide great cost savings over the current point-to-point architectures. A PON network is more fault-tolerant than an SDH-ring. This is supported by the fact that the PON node resides off the network, so that loss of power to a node does not affect any other node. This is not true for SDH, where each node performs regeneration.

Figure 7.12: Passive optical node vs active optical node

______66 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Because of its point-to-multipoint nature, a PON can broadcast video at a lower cost. As a result, PONs offer the most cost-effective solution for upgrading the critical last mile infrastructure to provide broadband capabilities.

7.2.1 ATM PON ATM PONs were developed in the mid-1990s through the work of the Full Service Access Network (FSAN) initiative. PON over ATM was a choice because ATM was considered to be suited for multiple protocols. Time Division Multiplexing is used in the downstream direction. The downstream traffic from the OLT to the ONUs is based on broadcasting a stream of ATM cells. An ONU picks up his own downstream traffic by matching the cells’ addresses to his own address. The upstream traffic from the ONUs to the OLT is based on a TDMA protocol. This protocol is necessary to avoid collisions of upstream traffic from multiple ONUs to the OLT. The differentiation of services is the nature of ATM, and this can be taken advantage of when differentiating between the services over the PON. However, the implementation costs are higher due to the significantly higher costs of the ATM components. The fixed cell structure of ATM also introduces an overhead “penalty” which yields lower operational efficiencies. The establishment of virtual connections and paths is rather time consuming, and this functionality is not necessary in the PON.

7.2.2 Ethernet PON Ethernet was specifically designed for carrying IP traffic. The advantages of Ethernet PONs are the following: • flexible service provisioning and rapid reconfiguration capabilities facilitate deployment • the ability to provision bandwidth in scalable 64 kbit/s increments up to 1 Gbit/s • the combination of IP and Ethernet simplifies the network configuration and reduces costs. • standard Ethernet interface eliminate the need for additional DSL or cable modems.

7.2.2.1 Differences between Ethernet and ATM PONs The key difference between Ethernet and ATM PONs is that in Ethernet PONs, data is transmitted in variable-length packets of up to 1,518 bytes according to the IEEE 802.3 protocol for Ethernet, while in ATM PON data is transmitted in fixed-length 53-byte cells as specified by the ATM protocol. For an ATM PON to carry IP traffic the packets must be broken into 48-byte segments with a 5-byte header each. This process is time consuming and complicated and adds additional cost to the OLT and ONUs. In contrast, Ethernet was specifically designed for carrying IP traffic, and therefore has a reduced overhead relative to ATM. Ethernet PONs offers higher bandwidth, lower costs and broader service capabilities than ATM PON.

7.2.2.2 Support and compatibility with the switch interface V5.x. The V5x interface has been developed from the static multiplexer interface V5.1 through to the dynamic multiplexer interface V5.2. The dynamic multiplexer interface V5.2 means that there is no traffic channel allocation for permanent customers, each customer can occupy any unoccupied channel. One interface V5.2 can consist of a maximum 16 links E1 (16 links of 2 Mbit/s), which means a maximum of 480 traffic channels. The blocking probability will depend upon the concentration factor applied.

______67 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Types of services that can be passed through interface V5.2 are customers of regular telephone (POTS), ISDN BA, ISDN PRA, and PABX DID.

Security aspects Multi-layered security such as VLAN closed user groups and support for VPN, IPSec and tunnelling is used for security purposes.

Eavesdropping the fibre Beams of light propagate through fibre cables by reflection on the insides of the cable. The inner surface will no longer be totally reflecting if the fibre cable bends. This leads to radiated power from the cable, which can be detected. It is therefore possible to eavesdrop optical fibre by bending it. This can be avoided by the use of encryption and restricting physical access to the fibre. If the traffic is especially confidential, the fibre cable used for the transmission should be equipped with an absorbing lining. This will complicate eavesdropping because the cable will then lose very little or even no power when bending. Supervision of the signal is also an option, to detect any alteration of the power.

Range vs bandwidth and number of connection points

1. Bandwidth: 1.25 Gbit/s Range: 20 km Number of connection points: 32 Bandwidth per customer: 39 Mbit/s

2. Bandwidth: 622 Mbit/s Range: Number of connection points: 32 Bandwidth per customer: 20 Mbit/s

3. Bandwidth: 155 Mbit/s Range: Number of connection points: 32 Bandwidth per customer: 4.8 Mbit/s

______68 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

7.3 Wireless Access Networks

Figure 7.13: The positioning of wireless radio access networks

7.3.1 Wireless Radio Access Networks Wireless radio access can be divided into two groups of systems. • Point to Multi-Point (PMP) • Point to Point Microwave links

7.3.1.1 Point to Multi-Point Nowadays the PMP Radio systems are a full alternative for accessing customers and are also an alternative to xDSL technologies. The available bandwidth per access is between 64 kbit/s and 8 Mbit/s. For higher bit rates it makes more sense to use conventional point-to-point microwave links. The system consists of a base station located on a high building with up to 4 sector antennas, each covering a 90° area. The end-user has a small 26 cm antenna on his roof and a small 19” indoor unit.

The advantages of PMP are: • Fast installation • Simple to install (smaller and simpler than a satellite dish) • Capacities up to 8 Mb/s • Types of services can be mixed

______69 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

One of the conditions for being able to use this access is that there is Line Of Sight (LoS) between the base- station and the customer. This also applies to Point-to Point links.

. Figure 7.14: Wireless Radio Access Systems

The following services are possible using PMP: • Connectivity up to 2 Mb/s • ISDN • Internet/Intranet • LAN-LAN Connections

All services can be mixed (data, leased line, speech) up to a total capacity of 8 Mbit/s per terminal.

______70 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Figure 7.15: Wireless Radio Access Services

7.3.1.2 Point to Point microwave radio links Point-to-Point Radio systems are a full alternative for accessing customers. The available bandwidths per link are between 2 x 2 Mbit/s and 155 Mbit/s. The frequency ranges vary from 38 GHz (short haul) down to 2.4 GHz (long haul). The requirement for using a microwave link is of course that Line of Sight (LoS) is available. Although in many cases, where direct LoS is not available, a passive or active relay can be used.

Transmission systems are generally characterised by their quality and their availability. The relevant parameters for digital transmission systems are the bit error ratio (BER) and the outage time. The BER is defined as the ratio of the total bits transmitted to the received errors within a certain period of time. Due to economic and physical constraints quality and availability can be realised only to a certain extent. The CCITT and ITU therefore recommend practical objectives for the digital radio link.

Some relevant aspects to be considered are:

Flat and selective fading Propagation effects that cause fading mainly influence the quality and availability of radio relay systems. Fading can be divided into flat and selective fading. Flat fading is caused by frequency-independent variations of the path attenuation due: • Above 12 GHz, to rain • Below 8 GHz, mainly to • multipath propagation with small time delays between direct and multipath signals ______71 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

• obstruction • diffraction • ducting • Between 8 and 12 GHz, all these effects have to be considered.

The influence of flat fading cannot be completely eliminated but it can be reduced to a great extent by optimising the antenna heights and increasing the flat-fade margin.

Selective fading means frequency-dependent variations of the path attenuation. It is caused by multipath propagation. The direct and indirect signals add in the receiver and produce variations of the amplitude and phase of the resultant receive signal with respect to frequency. Due to the sensitivity of a digital radio relay system such degradations can cause bit errors and even system outages. Apart from conservative route planning, frequency and space diversity, frequency and time domain equalizers and error-correcting codes are the best counter measurements against selective fading

System loss and flat-fade margin When considering the error performance and availability of a radio link, system loss and flat-fade margin are determined. They are the basic parameters.

The system loss of a radio link is defined by the sum of all attenuations between transmitter output and receiver input. Considered are: • feeder losses • branching losses • additional losses (attenuation due to obstructions, passive repeater(s), attenuators to reduce transmitted power) • free space loss • antenna gains. The transmit level minus the system loss gives the receive level. The flat-fade margin is defined as the difference between system gain and system loss, or between the un- faded level at the receiver input and the receiver threshold. The flat-fade margin is calculated for each hop of a radio link on the basis of the length of the hop and the climatic and topographic conditions, in order to check that the ITU objectives for quality and availability will be met. Because up-fading can occur due to ducting and defraction, a margin for the maximum receiver level must also be considered. Normally 6 dB are sufficient for defraction, but 15 dB or more must be allowed if severe ducting is expected.

Performance The error performance calculation is the short-term quality prediction for a digital radio relay system. The ITU has therefore defined the parameters "degraded minutes" (DM) and "severely errored seconds" (SES). The result of the calculation provides the occurrence probability of DM and SES during any month. The total outage probability is determined by the sum of the separately calculated: ______72 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

• Outage probability due to flat fading, • Outage probability due to selective fading and the • Increased flat-fading outage probability due to RF interference. The outage probability due to flat fading is calculated by the method taken from ITU Rep. 338-6, 2.3.1.1. The outage probability due to selective fading is calculated by the method developed by G.K. Grünberger and M. Glauner in 1989 using a two-ray model and the signature method. The increase of flat fading-outage probability due to RF interference is calculated by determining the reduction of the flat-fade margin in case of interference. Of great importance for the performance calculation is the right choice of the climatic and topographic parameters. They should be proposed by the administration of the country involved. If reliable values are not available they can be taken from ITU Rep. 338-6, Annex II, Table III. For countries and regions, which are not represented in this table, the best compromise is to use the values proposed by the USA.

Non-availability due to Rain Radio systems operating above 10 GHz are affected by attenuation due to precipitation. Snow and hail are negligible with respect to rain. Rain attenuation is caused by the scattering and absorption of the radio ray by the raindrops. The attenuation increases in proportion to the intensity of the rainfall. Due to the fact that the shape of raindrops is not absolutely round the attenuation of the polarization planes is different. In general the attenuation of horizontal polarized waves is higher than of vertical polarized waves. The non-ideal shape of the raindrops and their inclined trajectory can also cause a decrease of polarization discrimination. Effective measures against rain attenuation are an adequate flat-fade margin or a reduced path length. Route diversity may be necessary in severe cases. The calculation method for outage prediction due to rain normally used is based on ITU Rep. 338-6, 2.4. Important for the validity of this calculation are exact statistics of the occurrence probability of certain rainfall intensities in the region under consideration. If reliable rain statistics are not available then the values provided by ITU Rep. 563-4 have to be used.

7.3.2 Wireless Optical Networks The demand for broadband services is growing dramatically, and new broadband access technologies are developing. One of these technologies is Wireless Optical Networking (WON), which is a broadband access technology that improves upon the concept of free-space optics. Wireless Optical Networking is a combination of two distinctive technologies: free-space optics and telecommunications networking. A WON delivers high-bandwidth access over the air using invisible beams of light. It consists of interconnected free- space optical communication links. The network provides multiple routes to each building, rerouting around network faults and interconnectivity to other communications networks. A free-space optical link consists of two optical transceivers that are aligned to each other with a clear line-of- sight. The optical transceivers provide full duplex capability by means of a laser transmitter and a detector. The transceivers are usually mounted on building rooftops, but can also be mounted on building sides and behind glass windows. Free space optics enables very fast deployment of broadband services to buildings and is also cost- competitive. The wireless optical links can be deployed during a couple of hours. This is then cheaper and more flexible, making this solution more attractive in many cases. ______73 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

WON technology is designed for commercial buildings in proximity to a metropolitan fibre ring in dense urban centres or business park environments. The technology can also be used to connect blocks of flats to a fibre ring.

7.3.2.1 WON Architecture A Wireless Optical Network (WON) consists of two optical (infrared) transceivers that are aligned to each other with a clear line-of-sight. The optical transceivers provide full duplex capability by means of a laser transmitter and a detector. The transceivers are usually mounted on building rooftops, but can also be mounted on the sides of buildings and behind glass windows. Another advantage is that infrared requires no licensing or right of way permits of any kind. WONs enable very fast deployment of broadband services to buildings and are also cost-competitive. Connecting buildings that are situated close to a fibre network can otherwise be time consuming and expensive. In contrast, wireless optical links can be deployed within 2 hours. Infrared offers a much higher bandwidth capability than either spread spectrum or microwave. In addition, infrared provides a more secure method of data transmission, by using a very narrow beam, which is very difficult to intercept. In order to intercept the transmission, the exact location of the beam as well as the ability to enter the direct path of the beam would be required. Also, any interception would be immediately detected because communication would be interrupted. Link encryption technology can be added to provide an even higher level of security. WON technology is designed for commercial buildings in proximity to a metropolitan fibre ring in dense urban centres or business park environment. The technology can also be used to connect blocks of flats to a fibre ring. WONs are very complementary to last mile technologies like fibre, microwave and copper. These technologies address different market segments based on the technologies, the technical capabilities like reach and bandwidth, and economic realities.

Figure 7.16: Example of a WON access infrastructure

Availability is determined by the length of the link and by the fog patterns in the location area. Shorter links lead to better performance, and the length of the links must be adjusted to the specific location area, given the weather conditions. Fog and sun affect the link, but rain does not present a huge problem for the optical link. ______74 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

To prevent damage from direct sunlight, filters can be built into the system. It is recommended, however, that the systems not be mounted in the East-West orientation to avoid direct sunlight. The optical transceivers have to take building movement into account, by accepting departures of the localisation of the beam. Where absolute availability is needed, a backup hybrid optical/microwave link can be installed, or a Radio Frequency wireless backup system can be built-in to mitigate potential downtime caused by weather or building sway. Shorter links and a redundant mesh topology can also be used to enhance the availability.

Range vs attenuation and availability 1. Range: 1 km Atmospheric attenuation: 20 dB Availability: 99 %

2. Range: 2 km Atmospheric attenuation: 13 dB Availability: 98 %

Summary:

Band- Range BER MTBF Avail- Price Topographic area Protocols width ability of application 1 Mbit/s - 200 m - 10-12 - 8 - 10 yrs 98 % - $50.000 Downtown urban MIBs for: 2.5 Gbit/s 3.75 km 10-9 per system 99.999 % per pair core . ATM Larger business . SONET campuses . Ethernet 802.3 Blocks of flats . IP over-Ethernet.

Table 7.1: Characteristics of WON

Network architectures There are several network architectures in which wireless optical networks can be deployed. One of the architectures is the point-to-point topology. In this solution, there is a point-to-point link between the sender and the receiver, and there are no alternative ways in the network for these devices to communicate. The advantages of this topology are that each link is independent, the architecture is very simple, and it does not require a lot of planning in advance. The disadvantages are that the optical link is a single point of failure, and that longer distances lead to lower availability.

Figure 7.17.1: WON (point-to-point architecture)

______75 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Another network architecture is the hub and spoke topology. In this architecture, several nodes are connected to a hub by means of free-space point-to-point optical links. This architecture offers the advantage that traffic is collected at a single point, which leads to an efficient connection to the core network. The disadvantages are again that each optical link is a point of failure and the availability decreases with longer link distances. The location of the Hub is critical to maximise the number of buildings with line of sight. Hub development and cost are also among the disadvantages of this topology.

Figure 7.17.2: WON (hub and spoke architecture)

The ring architecture is yet another topology. This architecture offers the advantage that each optical link is not a point of failure because the node can be reached from two directions. The optical links also tend to be shorter than in the earlier scenarios,. Traffic is collected at specific points, which gives efficient connection to core network. The disadvantage with this approach is that the traffic is sent in both directions, which wastes over-the-air bandwidth. The network architecture is also more complex. Large number of over-the-air hops or multiple overlapping rings leads to poor scalability for large network.

Figure 7.17.3: WON (ring architecture)

The mesh architecture is a very common topology for wireless optical networks, and this topology is the most reliable one. In this architecture the nodes are meshed together, which results in each node having several alternative ways to reach all the other nodes in the network. A few of the nodes are connected to the core network. This approach has the following advantages: the optical link is not a single point of failure because mesh allows alternate routing. Shorter links and mesh re-routing give higher availability. Traffic is collected at a specific point, giving efficient connection to core network. One disadvantage is that the mesh architecture is more costly, as multiple links are needed for each building. Also, the architecture is complex and requires planning in advance. ______76 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Figure 7.17.4: WON (mesh architecture)

Future development of WON technology will include the introduction of a WDM layer.

7.3.3 LMDS The Local Multipoint Distribution Service (LMDS) is a broadband wireless point-to-multipoint communication system operating above 20 GHz (the precise frequency band depends upon country of licensing) that can be used for the provision of reliable digital two-way voice, data, Internet and video services. The term Local indicates that the propagation characteristics of signals in this frequency range limit the potential coverage area to a single cell site (transmission range up to 5 kms, in metropolitan areas). The term Multipoint indicates a broadcast signal to the subscribers (while the return path is a point-to-point transmission). Distribution stands for the distribution of signals, which may consist of simultaneous voice, data, Internet and video traffic. LMDS provides high capacity point to multipoint data access that is less investment intensive than a wireline solution, is faster to deploy and is able to offer a combination of applications. The advantages of using this fixed wireless technology are as follows: • Lower entry and setup costs • Easy and fast deployment (all the equipment can be carried and installed with great ease) • Demand-based installation investment, which is a major advantage, compared to the wired architectures. Equipment only needs to be installed after a customer signs up for the service. • Cost of upgrading can be substantially less, as there is no other infrastructure than the end equipment. • There is less overhead of charging the transmission equipment and many problems of wired LANs such as tracing of damage in transmission equipment, do not exist at all. • Once the basic infrastructure is aligned, QoS can be achieved. • Bandwidth reuse is very high because of the cell structure used. • Network management, maintenance and operation costs can be very low.

______77 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

LMDS operates at 40 GHz in Europe and was first initiated with the establishment of Digital Video Broadcasting (DVB) by the European Broadcasting Union. The technical specifications given by DVB became the ETSI standards. Further specifications were added by the Digital Audio Video Council (DAVIC) regarding the return channel protocol and modulation methods to be adopted. Other standard activities currently underway include activities by the ATM Forum and ITU. The majority of these methods use ATM cells as the primary transport mechanism.

7.3.3.1 LMDS Network Architecture A normal LMDS setup has a central facility with access to the PSTN and Internet via a fibre-link. The users are then accessed by a relay signal over point-to-point microwave links, which in turn pass the signal along the hubs, located on rooftops or as stand alone towers, for the point to multipoint transport to the end site. The LMDS network architecture consists of primarily four parts: 1. Network operations centre (NOC) 2. fibre-based infrastructure 3. Base station 4. Customer premise equipment The network management equipment for managing regions of customer network come under the NOC. Multiple NOCs can be interconnected. The fibre-based architecture consists of SONET OC-12, OC-3 and DS- 3 links, ATM and IP switching systems, the local exchange equipment and interconnections with the Internet and PSTN networks. The base station is where the conversion from fibred infrastructure to wireless infrastructure occurs. Base station equipment includes the network interface for fibre termination, modulation and demodulation functions, microwave transmission and reception equipment. Local switching can also be present in the base station. The customer premise equipment varies widely from vendor to vendor. All configurations include in-house digital equipment, modulation and externally-mounted microwave equipment. The customer premises equipment may attach to the network using TDMA, FDMA or CDMA. Different customer premises equipment require different configurations. Customer premises equipment that is supported includesDS0, POTS, 10BaseT, Unstructured DS1, structured DS1, frame relay, ATM25, serial ATM over T1, DS-3, OC-3 and OC-1. The customer premises locations can range from large enterprises to mall locations and residences.

7.3.3.2 System architecture LMDS system operators offer different services and as a result the system architecture differs. The most common architectural type uses co-site base station equipment. The in-house digital equipment connects to the network infrastructure and the external microwave equipment mounted on the rooftop is housed at the same location. Typically the radio frequency planning for these networks uses multiple sector microwave systems, in which transmit and receive sector antennas provide service over a 90, 45, 30, 22.5 or 15 degree beamwidth. LMDS combines high capacity radio-based communications and broadcast systems with interactivity operated at millimeter frequencies. Interactive LMDS has a point-to-multipoint downlink and a point-to-point uplink (see Figure 7.18).

______78 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Microwave Tower

Building 1

Transmitter Site

User Locations

Figure 7.18: LMDS for broadcast and interactive services

The transmitter site should be on top of a tall building or on a high pole overlooking the service area. The transmitter covers a sector typically 60-900 wide. Full coverage thus requires 4-6 transmitters. The streams transmitted contain 34-38 Mbit/s of data addressed to everybody in the coverage zone, subgroups or individuals. The capacity of the point to point return channels is determined by the needs of the individual user. Operation of LMDS in an area will normally require a cluster of cells with separate base stations for co- located transmitter/receiver sites. One of the base station sites will serve as the co-ordination centre for the franchise area and connect the LMDS cells to external networks. Intercell networking may be implemented using fibre or short hop radio relay connections.

Licensing and deployment in Europe indicate that there will be systems in different frequency bands from 24 GHz up to 43.5 GHz. Already the frequency band 24.5-26.6 GHz with sub-bands of 56 MHz has been opened for point-to-multipoint applications in many European countries.

7.3.3.3 Technologies employed In LMDS, the high capacity broadcast-based downlink is shared among several users in a flexible way. The front-end technology is still expensive at millimeter frequencies but existing high electron mobility transistor modules offer the required performance. The front-end technology at 40GHz is more expensive than at 28-29 GHz and attenuation by precipitation increases with frequency, favoring the lower frequency ranges. The higher capacity offered at 40GHz may compensate for these effects in the long run. The number of transport streams is determined by demand and limitations set by available spectrum. This gives a scalable architecture, starting with relatively low capacity and adding transmitter modules as demand increases. The transmission format for Digital Video Broadcasting (DVB) satellite transmission based on quadrature phase shift keying (QPSK) modulation has been adopted by both DAVIC and DVB and with the same IF interface, 950-2150MHz, between outdoor and indoor units. In DVB IP or ATM data are included in the

______79 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

MPEG transport stream in combination with TV programs while DAVIC has separate high-capacity ATM based data transmissions. The uplink is the individual connection and different technologies may be used depending on demand. Two of the broadband driving applications, interactive TV and Internet will require only low-capacity return links and technologies like GPRS and PSTN/ISDN. For more demanding applications an in-band radio return link with on-demand capacity is required. The total capacity of a system is mainly determined by the available frequency resource. In a cellular system employing QPSK modulation, the capacity of a 2 GHz system is easily 1.5 Gbit/s per cell for downlink and uplink.

Type Uplink data rate Downlink data rate Max range (km) LMDS 0-8 Mbit/s typical 36Mbit/s (shared) 5 25.8 Mbit/s possible

Table 7.3: LMDS capacity features

7.3.3.4 Protocol Layer Most of the available commercial LMDS systems provide Ipv4 services to end users using IP over ATM. The use of an ATM platform allows co-operation between radio systems and ATM core networks. However, the use of ATM produces a large protocol overhead with respect to the payload information carried by the IP packets. In order to optimise transmission of IP connectivity, IP packets can be transmitted directly avoiding the ATM layer. Moreover, the radio data link control (DLC) and medium access control (MAC) protocols should be optimised for IPv4/ IPv6 characteristics.

7.3.3.5 Applications (Services) LMDS is first of all a system of high flexibility allowing for capacity on demand. Changing the cell size through reduction of either cell diameter or illumination angle increases total capacity. It’s flexibility with regard to high on-demand capacity in both directions makes it well suited to home offices and users in local domain. The first major applications are TV and Internet and business oriented, thus combining professional and entertainment use. It is considered a supplement / alternative to cable TV and new broadband applications in areas with a certain population density. It is expected that availability of high-capacity access to every home will greatly stimulate the development of the information society and new broadband applications. Emerging applications are probably also teleworking and wireless video surveillance. Also, the Telephony service could be possible (VoIP). LMDS is best suited for cases where operators are required to quickly deploy data communication capability in high population density areas, or when they do not have access to copper infrastructure. Another clear opportunity is to deploy services over LMDS into low population density areas, where a fixed copper or fibre infrastructure might be too expensive or static.

7.4 Wavelength Division Multiplexing (WDM) Wavelength Division Multiplexing is an optical multiplexing technique used to increase the carrying capacity of fibre and wireless networks beyond what can currently be accomplished by Time Division Multiplexing (TDM) techniques. WDM has been used in the backbone for quite some time, and is now starting to be ______80 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers introduced for access networks. High bandwidth, enhanced flexibility and upgradability are the main advantages of WDM.

7.4.1 WDM in a PON Further cost reductions can then be achieved in a PON through the addition of a WDM layer. This addition is easy to accomplish in a PON because the nodes sit “off” the backbone. Different wavelengths of light are used to transmit multiple streams of information along a single fibre with minimal interference. WDM introduces increase in capacity and protocol transparency.

7.4.2 WDM in access networks One differentiating characteristic of an access network is that it directly interfaces with the customer premises and thus must accommodate a range of data formats and data rates. An access network must be scalable, in terms of both the number of customers and the demands of any given customer. In a WDM solution, additional customers can be accommodated by adding more wavelengths (up to a limit). WDM provides graceful upgrades, as assigning a wavelength per customer naturally isolates the customers to a large degree. The flexibility and scalability of a WDM access architecture shows a potential in deploying access networks that can meet the future customer demand in both the business and residential markets.

7.4.3 Application areas of WDM Increase of bandwidth in the infrastructure is the basic application. WDM offers more capacity through the installation of overlay equipment on the shared fibre infrastructure.

Figure 7.19: Infrastructure capacity upgrade ______81 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Segregation of services Another area of application is segregation of services. This is realised by offering different services over different wavelengths.

Host Television λ 1 Fibre λ 1 1

Router Router Fiber optic terminal Fiber optic terminal IBM PS/2 C V Satellite dish

Disk array Public switch

Satellite dish

Television λ 1 λ 1 λ λ 2 W 2 λ W 1 Fibre 3 D D λ Host 3 M M IBM PS/2 λ 4 λ C 4 Fiber optic terminal V Fiber optic terminal

Public switch

Disk array Figure 7.20: Service separation

Segregation of customers Customer segregation is yet another application area for WDM. The technology allows independent overlay networks to share the same fibre infrastructure, and differentiates customer per gross bit rate used as well as by the quality of service required.

______82 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

ONU 1- Residential

ONU 2 - Residential

WDM 0

& 0

0

0 0 &

0 0 & WDM 0

0 WDM ONU3 - Residential OLT

WDM

0

0

& 0

Router ONU 4 - Residential

ONU5 - Business

Figure 7.21: Customer segregation

Selecting of service provider From a user perspective, the freedom of selecting service provider may be regarded as a service. A connection to a certain service provider could be performed on the optical layer by assigning one or more dedicated wavelengths between the customer and the desired service provider. This requires a WDM network, and the technology used to realise the service is Frequency Division Multiple Access (FDMA).

Figure 7.22: One or more ONUs can be linked to one or more service providers

______83 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

WDM Comparison survey (for PONs)

Band- Band- Range Number of Avail- Price Area of application Protocols width of the width connection ability system per user points 155 Mbit/s 5 – 30 20 km, 16 - 64 99.999 % 20% of SMEs, - ATM – 2.5 Gbit/s Mbit/s 9 – 12 SDH Residential - Ethernet miles subscribers, PON - Gigabit MANs, Ethernet Business campuses, - Fibre Interconnection of Channel buildings, - ESCON Metro edge - TDM applications - PSTN - IP 51 Mbit/s - 5 - 30 20 km 32 - 64 99.999 % SMEs, Downstream: 2.5 Gbit/s Mbit/s Residential - TDM subscribers, WDM Upstream: MANs, PON Normal - TDMA bitrate: Business campuses, 155 – 622 Interconnection of Mbit/s buildings, Metro edge applications

Table 7.4: Characteristics of PON and WDM PON

______84 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

8. The Local Access Point

The Local Access Point (LAP) is a crucial part of the TORRENT architecture as it is equivalent to the local exchange. The LAP is therefore an access network termination device with enhanced functionality for the appropriate forwarding of traffic, eg. depending upon the type of requested service, the load situation of the available backbone networks, the currently generated costs, forwarding constraints issued by customer or operator/provider and other criteria. The availability of certain services will require additional mechanisms to be provided by the LAP like for example Video on Demand (VoD) or IP telephony may require caching or proxy/gateway functionality respectively. The LAP will also be the platform for hosting the agent system and also the management system being evident for the TORRENT architecture. All services which can be provided or implemented centrally will be located on the LAP as well. For that, the LAP is required to provide communication and configuration interfaces not only to the Residential Gateway (RG), but also to customers, operators and service providers in order to enable an integrated operation of services in the TORRENT architecture. Authentication, authorisation, and accounting (AAA) functions, firewalling, management of service level agreements (SLA), and network and service management functions are also located on the LAP.

Authentication Issues for Service Providers and Network Operators Service providers need to be concerned about providing customers with user-friendly secure authentication for access to services. Authentication becomes complicated as more and more services are provided over the network, especially as each service usually requires a separate password authentication. Users generally choose passwords that are easy to remember and often use the same password over and over, making it easy for an attacker to gain unauthorised access to services for the purposes of mis-use and fraud. To simplify the authentication process, service providers should provide strong authentication to services using smart card technology, certificates, and PKI-based (public key infrastructure) methods. Public key cryptography and a public key infrastructure should be used for scalability, eg. key management, and to provide digital signatures. Ideally, in the future, a user should have one smart card containing a certificate that is linked to the user’s service profile. Using such a card, a user can log on to the network and obtain access to all the services that the user subscribes to, instead having to maintain a different password for each service. It is feasible that a Radius Server maintains the user profiles so that the user is authenticated based on the secret key in the certificate without having to enter a password for each service. The smart card products can also provide the user with the possibility for making secure transactions with confidentiality and digital signature eg. secure payments over the Internet.

8.1 The LAP hardware architecture in the TORRENT project The Local Access Point will be designed around an architecture for converged networks, combining a Flextel switch with special purpose hardware proposed for the HIAD system: WebVision 4012 The Flextel switch is based on a backplane architecture with up to 12 slots providing different types of busses: • One interprocessor busses (IPB) with 2.1 Gbit/s of throughput • An I/O bus for connecting up to 22 market available ISA or PCI in two separated subsystems • A TDM bus for real-time data • Two manageability busses for exchanging system information between processors and I/O cards ______85 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

The backplane can be configured into subsystems for different purposes. The architecture of the backplane is shown in Figure 8.1.

Interprocessor Main bus 1 (2.1Gb/s)

Interprocessor Main bus 2 (2.1Gb/s) I/O Bus TDM bus Two Manageability buses

Electronical Switches for Backplane segmentation

Figure 8.1: Flextel switch backplane architecture

The chassis consists of the backplane and hot swappable components including: processor modules, power supplies and the clock control module. Also I/O module can be hot plug/unplug, but if the market card hosted is not hot swappable the functionality is useless).

All other components of the switch are available as modules to the chassis: • Processor module • Double or mono PIII processor with up to 850 MHz • Up to 1 Gbyte of main memory • Optional silicon or EIDE disk • Optional PCI card or CD/floppy drive • VGA and Ethernet controller • Keyboard/Mouse/USB • Manageability microcontroller • I/O module • Two standards PCI or AT slots • TDM bus extensions • Manageability processor • Hot swap capabilities • SCSI disk module • SCSI U2W disk interface ______86 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

• Two hot swappable disks • Optional RAID 0/1 • SCSI disk array module • SCSI U2W disk interface • Up to 6 hot swappable disks • RAID 0/1/5 feature The backplane segmentation feature in combination with processor and I/O modules allows for building subsystems for different purposes, eg. for running a firewall, a cache, a network management platform or other software. Different operating systems (LINUX, Windows NT/2000) can be hosted on different processor modules. The usage of market available standards PCI cards allows the provision of a variety of physical interfaces to connect to different types of networks or other additional equipment. A management architecture exists for the Flextel switch, which is based on the manageability busses, the management controllers on the different modules and a management software running on at least one of the processor modules. A local API allows for configuration and event management either locally from a generic application or remotely via a standard SNMP manager or a web browser. The remote management needs also a local SNMP agent in combination with the private proprietary Flextel MIB (plus probably also standard and private MIBs for protocol stacks and extension cards) interfacing to the local API. This management software architecture is shown in Figure 8.2.

SNMP Manager WEB browser

Local WEB Application Agent

SNMP Agent MIB

Flextel Management API

CS&M

HOT SWAP COM DRV DRV & TOOLS

Figure 8.2: Flextel switch management software architecture The LAP will be realised as an extension to the Flextel switch. A number of new, custom communications modules will provide the required functionality of an access network termination. The back-end of these modules (meaning the communication to the switch) is realised in the same way as on the available standard modules, and the front-end (meaning the interfaces to the outside) has to be developed according to the requirements. These modules then have to be integrated into the Flextel management software architecture. The design will use interface cards to host the modem chipsets for the line termination, a traffic concentrator, facilities for processing the data streams and an interface to the controller which provides communication and access to the available bus systems on the back-plane.

______87 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

The different communications modules to be attached to the Flextel system, and that together will constitute the HIAD system, will be selected from the following:

1.- xDSL line card This card aggregates the traffic of up to 32 DSL (ADSL, VDSL, ISDN) users over the back plane. The xDSL card will include the Inverse Multiplexing over ATM capability.

2.- Cable line card. The cable line card will support 1 27Mbit/s downstream channel and up to 8 downstream channels (supporting several bit rates depending on the quality of the link). The average number of users that may be service provisioned with this card for an average of 80kbit/s per user is of about 350, this may increase depending on the targeted QoS.

3.- Fibre line card: The fibre line card will support the connection to the backbone network, this card will serve up to 4 fibre links that will be running under the IP or ATM protocol, the data rates that will be supported by each of this links are DS3 (45Mbit/s), STM-1 (155Mbit/s) and STM-4 (622Mbit/s)

The architecture of these 3 cards is generic and it will mainly vary in the chipsets interfacing to the physical media, the following diagram shows a preliminary schema that will be followed:

ATM bus. H W

& PHY line PCI bus Interface: S W - xDSL. - Cable card. I - Fiber optics. N TDM bus T E R F A Management C bus. E

Figure 8.3 The generic schematics of the line cards in the HIAD system.

4.- Core-switching ATM card: The core-switching card is the heart of the HIAD system and it is the switching fabric that will manage the traffic flowing across the LAP from the subscribers to the backbone and vice versa. This core card will manage each ATM connection that passes through the LAP and particularly will enable service provisioning and queue management on a per VC basis (SVC and PVC).

______88 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

The core-switching card has a different schema than that exposed for the line cards, this can be seen in Figure 8.4:

Figure 8.4: The architecture for the core card in the HIAD system.

5.- The Public Switched Telephone Network interface card: The PSTN card is the appropriate media through which TDM traffic and packet will get the possibility to converge in the HIAD architecture, the card will have the following functionalities: • VoIP gateway. • V5.2/SS7 interface card. • Virtual pool modem.

The card has the schema suggested above for the line cards; the main differences include the physical chipset, and the possibility to include a daughter card in order to perform digital signal processing functions. The preliminary version will provide 8 E1/T1 lines trunks to the PSTN/TDM backbone.

8.2 The LAP software architecture in the TORRENT project It is necessary to differentiate between the base technology and the protocol structure. The technology software base are the tools and operating systems involved in the design of the LAP, whereas the protocol structure incorporates different software stacks that are integrated together above the technology base and that constitute the main added value in the LAP. ______89 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

TORRENT will be decomposed in two trial stages at different time schedules within the duration of the programme. Theses two stages will enable the testing of different protocol architectures that are described below:

8.2.1 The TORRENT software structure in the first trials.

Layer Residential Gateway Local Access Point Layer

Home Network Access Access Core Network Network Network

Thin client Storage and architecture processing Streaming 4 TCP, H.323, TCP, H.323, 4 SIP. SIP. 3 IP IP IP IP, V5.2 3 2 802 ATM (*) ATM (*) ATM 2 1 10/100 ADSL, Copper, IMA, ADSL, Fibre, Coaxial 1 Ethernet IMA(**) IMA(**)

Table 8.1: The protocol stack for the first trials

(*) ATM will be used to encapsulate IP traffic over xDSL. Cable networks are already based on IP.

(**) IMA refers to inverse multiplexing over ATM and enables the capability of emulating fibre like capacities over multiple copper pairs.

8.2.2 The TORRENT software structure in the second trials. The second stage of the TORRENT infrastructure will likely include the following additional issues: Physical infrastructure: 1.- Full cable CMTS DOCSIS 1.1 capabilities (part of this development maybe ready for the first trials) 2.- Inclusion of VDSL capabilities. 3.- ISDN line card available. Software infrastructure: 1.- VoIP Gateway (Media Gateway). 2.- SS7/IP gateway. 3.- MPLS/RSVP. 4.- Full application service provisioning management system (ASP). 5.- Final election for the caching system. ______90 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

ATM delivers important advantages over existing LAN and WAN technologies, including the promise of scalable bandwidths at unprecedented price and performance points and Quality of Service (QoS) guarantees, which facilitate new classes of applications such as multimedia. These benefits, however, come at a price. Contrary to common misconceptions, ATM is a very complex technology, perhaps the most complex ever developed by the networking industry. While the structure of ATM cells and cell switching do facilitate the development of hardware intensive, high performance ATM switches, the deployment of ATM networks requires the overlay of a highly complex, software intensive, protocol infrastructure. This infrastructure is required to both allow individual ATM switches to be linked into a network, and for such networks to internetwork with the vast installed base of existing local and wide area networks. An ATM network consists of a set of ATM switches interconnected by point-to-point ATM links or interfaces. ATM switches support two kinds of interfaces: user-network interfaces (UNI) and network-node interfaces (NNI). UNI connect ATM end-systems (hosts, routers, etc.) to an ATM switch, while an NNI may be imprecisely defined as an interface connecting two ATM switches together. Slightly different cell formats are defined across the UNI and NNI. More precisely, however, an NNI is any physical or logical link across which two ATM switches exchange the NNI protocol. ATM circuits are of two types: virtual paths, identified by virtual path identifiers (VPI) and virtual channels, identified by the combination of a VPI and a virtual channel identifier (VCI). A virtual path is a bundle of virtual channels, all of which are switched transparently across the ATM network on the basis of the common VPI. All VCI and VPI, however, have only local significance across a particular link, and are remapped, as appropriate, at each switch. ATM networks are fundamentally connection oriented. This means that a virtual circuit needs to be set up across the ATM network prior to any data transfer. The fact that ATM is connection oriented implies the need for ATM specific signaling protocols and addressing structures, as well as protocols to route ATM connection requests across the ATM network. These ATM protocols, in turn, influence the manner in which existing higher layer protocols can operate over ATM networks. The basic operation of an ATM switch is very simple: to receive a cell across a link on a known VCI or VPI value, to look up the connection value in a local translation table to determine the outgoing port (or ports) of the connection and the new VPI/VCI value of the connection on that link and to then retransmit the cell on that outgoing link with the appropriate connection identifiers. One of the great advantages of ATM is its support for guaranteed QoS in connections. Hence, a node requesting a connection set up can request a certain QoS from the network and can be assured that the network will deliver that QoS for the life of the connection. Such connections are categorised into various types of ATM QoS types: CBR, VBR, ABR, and UBR, depending upon the nature of the QoS guarantee desired and the characteristics of the expected traffic types. Depending upon the type of ATM service requested, the network is expected to deliver guarantees on the particular mix of QoS elements that are specified at the connection set-up (such as cell loss ratio, cell delay, and cell delay variation).

______91 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

9. Core Network Protocols, Interfaces and Functionalities

The main functions/requirements imposed on the core/backbone network are: • High speed switching/routing and transport • QoS interworking • Protocol independence • Scalability • Dynamic bandwidth management to allow on demand provisioning

9.1 Carrier Networks At present, most of the carrier core networks use PDH, SDH, ATM and Frame Relay (FR) as transport technologies. The trend is to consolidate the network core onto just an ATM/IP layer, and in order to accomplish it; FR to ATM inter-working (ITU-T Recommendation I.555) might be implemented at the network edge. In this way FR would be smoothly moved out of the core network towards the network edge (where FR demand is still growing accordingly to various market studies) acting as a pure access bearer (some equipment/cards re-use should be also possible). 1.- ATM interfaces: With capabilities ranging from STM-1 to STM-4 depending upon the number of users that are being serviced by a given exchange. Many emerging broadband access networks (eg. ADSL) have ATM interfaces, which can be a good argument for using also ATM in the core. The Frame Based ATM over SONET/SDH Transport specification for main rates (SDH standard for transmission over OC-3 optical fibre at 155 Mbit/s, STM-4 at 622 Mbit/s, STM- 16 at 2.5 Gbit/s, etc) are described in the ATM Forum (AF-FBATM-0151.000). 2.- IP interfaces: ATM to the exchange, however, is facing tough competition from the IP protocol, due to the proliferation of IP-based terminals. Transporting IP over ATM adds a high overhead (10%) associated with ATM headers. IP is one protocol on which services can converge, and therefore, by applying IP in the LAP would establish a straightforward way to bridge between an IP core Internet and IP-based access network technologies. Overhead would be reduced and management should be better unified.

In the case of DSL access networks (where ATM is used as a transport mechanism in the access network), an encapsulation of the IP protocol has to be done before sending the data to the customer. The additional overhead is unavoidable.

3.- TDM interfaces: TDM interfaces (eg. basic- and primary- rate ISDN) are used where the interfacing of time sensitive traffic to the backbone network (eg. STM) is necessary. In particular, for voice and video services and some preferred data applications. The SDH based transmission is standardised (G.957, G.958, G.703 and G.707/Y.1322) and on PDH (see recommendation G. 703, G.705 and G.751). ITU-T recommendation G.804 defines the ATM cell mapping on PDH. 3.1.- TDM/Voice services: Traditional TDM equipment is based on the regular switching of 64kbit/s timeslots. Though once perfectly suited to voice, this has been compressed to use increasingly lower bandwidths for the same quality (eg. GSM at 6.4kbit/s). The 64kbit/s TDM slots can therefore be subdivided to transport several circuits, but all circuits in a timeslot have to be

______92 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

switched to the same destination (network node). A more flexible use of the overall capacity of a TDM link, is to use packets. 3.1.- TDM/Data services: TDM data services are applied to cases in which the applications require a predictable network link; a feature that sometimes may not be provided by the traditional packet services.

Torrent's LAP will offer these 3 types of core network interfaces in order to attain maximum flexibility both for the physical installation and suitability of the equipment as well as a flexible provisioning system.

9.2 CATV Networks A cable network for broadcast applications may serve from a few thousand to a few million subscribers. The complete end-to-end system usually consists of a master (national) head-end, secondary (regional) head-ends, hubs, fibre nodes and digital receivers in the subscribers’ homes. The bi-directional data transfer progressively provided by cable networks is the key to a wealth of interactive services such as video on demand (VoD), near video on demand (NVoD), interactive TV, TV shopping and more. Content is received at the head-end as transport streams from a variety of satellite, terrestrial and telecoms network sources (SDH, ATM). Datacenter

Subscriber registration center

STB Internet application servers Regional streaming media Receiver Proxy servers servers MPEG encoder Billing system Regional streamning media Remultiplexers caches and modulators Satellite Streaming media Ad insertion manager

PSTN PSTN Satellite dish Off-air channels Signalling Gateway

Master Media Gateway Head-end Satellite Off-air Media Gateway dish channels Controller Internet

Secondary Secondary Head-end Head-end

Secondary Head-end Receiver STB Internet application servers MPEG encoder MPEG VoD server HFC Remultiplexers Network and modulators Web caches

Local streaming media servers

Local streamning media caches

Figure 8.5: The architecture of a CATV system. ______93 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

9.2.1 Broadcasted TV Broadcasted TV is a service that deals with the distribution of television signals to several (non addressed) customers, which is usually accomplished by off the air (terrestrial), satellite or cable (CATV) networks. Until the recent past, television was broadcasted in analogue systems, such as PAL or NTSC, but today many operators are moving to digital platforms, using the DVB standards. DVB uses MPEG2 to compress and transport video, audio and system, providing a global framework for enhanced TV services. Some additional features to basic TV service, like teletext, closed captions and conditional access, were already provided in analogue systems but there are several others that potentially may be supplied. Although the real-time nature of TV broadcast service has not change, user interactivity is being introduced. This implies the need for a return channel that does not exist on current CATV networks. Moreover, with the use of MPEG2, the bandwidth requirements have decreased by a factor of about 8, which leaves available bandwidth for additional channels or other services (eg. Internet access). The CATV service portfolio has migrated from just broadcast TV to the introduction of pay channels, then pay-per-view, NVoD, VoD, Internet access, and VoIP. Incremental introduction of new services has also been followed by an increasing personalisation of offerings, which is somewhat contradictory with the original non-addressed broadcast nature.

9.2.2 Voice over Cable (VoCable) Unlike copper pair networks, which were intended to support point-to-point communication, cable networks were originally designed to broadcast one signal to a large number of recipients, so there was no need to allocate bandwidth for individual subscribers. Therefore, to enable cable based IP telephony, modifications are required to the way that bandwidth is allocated and packets are delivered, whilst at the same time keeping a large amount of bandwidth available for the usual broadcast services. The first steps were taken with DOCSIS that defined the base rules for transmission and routing of packets over cable networks. DOCSIS v1.1 added QoS [CableLabs SP-RFIv1.1-I05-000714], such as packet prioritisation, and security features [CableLabs SP-BPI+-I05-000714] required for voice communications. Packet cable specifications, also developed by Cable Labs, have provided additional features like a protocol for signalling voice calls over cable networks, designated Network-based Call Signalling (NCS) [CableLabs PKT-SP-EC-MGCP-I03-010620]. This protocol uses network-based call agents to negotiate cable based IP telephony calls. NCS also supports IPsec as a mean to improve security aspects. Efforts to make US PacketCable specifications suitable for European networks are being made by EuroPacketCable. These include the interface requirements between IPCablecom domain (hybrid fibre-coax cable television network plus IP network managed portion) and other network domains. A gateway device will provide the appropriate interfaces among these networks. For interfacing with PSTN networks, the gateway should include a signalling gateway, a media gateway and a media gateway controller. The interface to IP networks will use protocols such as SIP, MGCP, MEGACO/H.248 and SIGTRAN. Interfaces specifications for PSTN, V5.2, ISDN, SS7, mobile networks and packet-based core networks will be developed.

______94 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

10. Mapping of Service Requirements to Network Resources

______95 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Home / Local Access Deployed in Service Quality * Bandwidth Remarks Network Residential / SME Telephony High bandwidth only possible over 64/128kbit/s DSL technologies have to be used for the Copper pair Internet short distances ISDN (high bandwidth) entertainment services Environmental 2-50Mbit/s DSL Terminals can be powered over the copper Entertainment pair

Residential / SME / Telephony Loss and delays associated with Up to 3Mbit/s If used in a shared media configuration, Ethernet Large Business Internet Ethernet are generally not important. throughput mechanisms are necessary to ensure that Environmental Loss is compensated for by higher priority is given to time critical services (30Mbit/s for Fast Entertainment level protocols Ethernet) Local powering in the terminal is required

Residential Telephony Ideal for Broadcast TV 100 TV channels Needs additional equipment to enable a Coax Internet return channel Up to 8Mbit/s Environmental downstream for Local powering in the terminal is required Entertainment Internet

Glass fibre may be Telephony Low loss, high bandwidth Limited only by Special tools required for the connectors Fibre (or plastic) used in large Internet the end makes glass fibre less attractive for businesses Environmental equipment residential deployment Entertainment Plastic fibre is a Local powering in the terminal is required cheaper alternative for residential use

Residential / SME / Telephony High bandwidth Up to 50Mbit/s Local powering in the terminal is required Firewire Large Business Internet Environmental Entertainment

Residential / SME / Telephony High bandwidth, short range Up to 11Mbit/s Local powering in the terminal is required WLAN IEEE 802.11b Large Business Internet Environmental Entertainment ______96 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Residential / SME Telephony Short range Up to 700kbit/s Licence-free, so may be subject to Bluetooth Internet (Asymmetric interference Environmental Data Rate Local powering in the terminal is required Forward)

______97 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Access Network Service Quality * Bandwidth Remarks Copper pair: Telephony Good quality voice on analogue lines 64/128kbit/s DSL technologies have to be used for the (high - Analogue telephony Internet High quality voice on ISDN ISDN bandwidth) entertainment services - ISDN Environmental High bandwidth only possible up to 1 2-50Mbit/s DSL CPE can be powered over the copper pair - xDSL Entertainment km Generally, all TV programmes have to be switched individually via the LAP. (Using VDSL, a few popular programmes could be broadcast, and the others (inc. pay-per-view) switched via the LAP)

Telephony Broadcast TV 100 TV channels All TV programmes can be transmitted Coax (CATV) Internet Internet voice Up to 8Mbit/s simultaneously Environmental downstream for Needs additional equipment to enable a return Entertainment Internet channel The upstream bandwidth over CATV depends upon the number of instantaneous users, and may not be sufficient for a video games service Local powering in the CPE is required

Telephony Losses and delays are associated with Up to 3Mbit/s All TV programmes can be transmitted Gbit/s Ethernet Internet an Ethernet link throughput simultaneously Environmental Loss is compensated for by higher (30Mbit/s for Fast If used in a shared media configuration, mechanisms Entertainment level protocols Ethernet) are necessary to ensure that priority is given to time critical services (eg. Telephony or Entertainment services) Local powering in the CPE is required

Telephony Low loss, high bandwidth Limited only by All TV programmes can be transmitted Fibre Internet the attached end simultaneously Environmental equipment Special tools required for the connectors makes glass Entertainment fibre less attractive for residential deployment Local powering in the CPE is required

______98 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Core Network Service Quality * Bandwidth Remarks Telephony Low loss Typically 155 ATM is generally carried over an SDH frame ATM Internet Low delay Mbit/s to 2.5 structure SDH Environmental Gbit/s Low jitter Entertainment

Telephony Low Loss Typically 2/34 TDM circuits are used to multiplex voice channels of TDM Internet Low delay Mbit/s 64kbit/s. they are often then further multiplexed into Environmental (1.5/45Mbit/s in SDH channels Low jitter Entertainment the US)

The transmission is over ATM or SDH links. Telephony MPLS offers a guarantee of quality Typically 155 Premium IP ** Internet (bandwidth, and delay) per defined Mbit/s to 2.5 (Internet) Telephony and Entertainment services need Environmental path. Gbit/s the low loss, low delay characteristics supported by Entertainment DiffServ offers only priority of ATM/STM or Premium IP over these technologies. Premium IP traffic over “normal” IP data

Telephony Losses and delays are associated with Typically 155 The transmission is over ATM or SDH links, but IP Internet a standard IP service. Mbit/s to 2.5 without any priority mechanisms, IP is not suitable Environmental Loss is compensated for by higher Gbit/s for Internet Telephony or Entertainment services Entertainment level protocols

Telephony Broadcast TV 100 TV channels All available TV programmes can be broadcast. Pay- CATV Internet Internet voice Up to 8Mbit/s per-view channels would be switched individually Environmental downstream for via the LAP Entertainment Internet Needs additional equipment to enable a return channel The upstream bandwidth over CATV depends upon the number of instantaneous users, and may not be sufficient for a video games service

Table 10.1: The Services that can be supported by different Home-, Access- and Core- Networks ______99 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

* Quality of Service is defined in ITU-T as the collective effect of service performance, which determines the degree of satisfaction of a user of the service. ** “Premium IP” = some form of support for QoS, such as one (or a combination of) the following: • DiffServ • MPLS • Over-provision of bandwidth

______100 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

11. Value-added features

11.1 Support for accounting The provision of accounting is an indispensable aspect of the provision of a service. It is the sole mechanism by which service providers and network operators get rewarded for the services they provide. With the increase in competition in the market place, providers are looking for better ways to maximise resource usage, while fairly apportioning charges for service usage. Traditionally services used simple charging schemes based on a flat rate charge. This is the situation depicted in the following figure.

Figure 11.1: Flat rate charging

This charging scheme may be potentially very unprofitable, if a large amount of users, use the service between 12PM and 3PM. Thus service providers are attempting to fairly apportion costs for service usage to the users. This requires more innovative and complex charging schemes to be developed. For example a Video Conferencing service might wish to use alternative charging methods based on bandwidth required, number of participants, security level, quality of service or any combination of these. Another source of complexity in charging schemes is cross-service discounting, where the usage of one service is heavily discounted when used in conjunction with another companion service. For example a premium IP service might be heavily discounted when used to download MP3 files from an online digital music store. In addition to the requirements that originate from user requirements, there are several requirements that the accounting system makes of service provider and network operators. These requirements derive from capabilities of the accounting system itself.

• Future-proof / extensible: Wherever possible, the accounting system must be future proof. As a bare minimum, the accounting system must allow new services to be deployed and rolled into the existing accounting system. In general, this is ensured by the following:

• Provision of open metering/charging interfaces: The metering interface is used for the collection of usage information. Clearly the use of a small set of standardised interfaces for the collection of usage data, reduces the need for complex mediation for each new service.

• Provision of flexible charging schemes: Clearly new services may require customised charging schemes, as services in general become more and more sophisticated. In most cases these charging schemes will make use of both network and service information in order to provide a bill. For

101 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

example, a digital library service may wish to charge customers more for downloading a large media file, as opposed to downloading a small article.

• Ability to perform online accounting: Traditional accounting systems usually have a paper-based approach to billing. However users are demanding more complex charging schemes, and more convenient access to the charging/billing information. To this end services must generate usage information

• Provision of open charging interfaces: The charging interface is used to collect pre-calculated charging information from other service or network providers. It gives the accounting system the ability to collect charges from other sources, and to merge / combine, in order to produce integrated bills

In today's liberalised telecommunication market, the network operator and service provider both have to attract customers while at the same time operating in a cost effective manner. Thus an important factor for both operators and providers is the ease with which a charging scheme can be set up, introduced to the customer and operated for the customer. The charging scheme must not cause network congestion or network instability. At the same time, the scheme should be profitable to both operator and provider. In what follows, we shall look at implementing and operating charging schemes and the infrastructure to support this activity.

11.1.1 Introducing New Charging Schemes The network operator and service provider will favour charging schemes that are easy to implement and easy to adapt to changing market needs [D9a][SCH1][SCH2]. The schemes should also work with a variety of transfer modes, services and service categories. Equipment and system manufacturers and infrastructure providers need to ensure that the network, servers and computing devices can provide appropriate support for charging schemes that the service providers and network operators deem suitable.

11.1.1.1 Architectural and Infrastrucure Considerationst Measurement Requirements: To support the operation of a charging scheme, the network and its associated computational nodes, should offer suitable facilities such as cell counting and call duration measurement. Cell counting will be limited by technology; counting at bit rates above 500 Mbit/s becoming challenging for semiconductor-based systems. Cell counts and duration information can be used to generate additional statistical information including peak and mean bandwidths. The assessment of the numerical aspects of QoS, such as cell loss and delay statistics will benefit from measurement facilities for these parameters. Integration: The scheme should be easily integrated into the existing environment consisting of the host computer and the network. Host Computer: The host computer should provide adequate software hooks into the operating system, and should facilitate data transfer to and from other facilities such as the MIB. The host computer should have sufficient memory to support the operation of the charging scheme. Other Interactions: Provision may have to be made to support charging schemes that need to interact with Connection Admittance Control, Usage Parameter Control and Network Parameter Control functions? Simplicity: The scheme should foster ease of understanding. The scheme should constrain programming complexity and consequently implementation errors. Mounting the software should be a quick and easy task.

102 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Will the charging scheme entail high customer care and customer billing costs because of its complexity and the need to educate the customer about the basis for charging? Expandability and Generality: It should be possible to expand, extend or modify to accommodate the changing aspirations and requirements of users, customers, network operators and service providers. Changes should be made easily and in a way that is understandable by, fair to, all interested parties. The range of ATM service categories to which an ATM charging scheme may be applied is spelt out by the ATM Forum which has identified the following categories: CBR, rt-VBR, nrt-VBR, ABR and UBR, to which the ITU has added the ABT category.

11.1.2 Technical Implications of Using a Charging Scheme A network operator will desire a charging scheme the use of which does not adversely influence network performance. Charging requires computation and bandwidth resources. Hence a charging scheme may reduce the bandwidth and computational resources available to the other traffic carried by the network and may also influence network performance [D9a][SCH1][SCH2].

11.1.2.1 Architectural and Infrastrucure Considerationst Computation may be needed to obtain usage parameters such as time and duration of connection, and mean and peak cell rate; computation may work with the Usage Parameter Control function; a considerable part of the computation may have to be on-line/real-time. Apart from time considerations, code and data sizes related to the computation or charging algorithm may need to be considered. For example, the Dynamic Memory requirements associated with a given charging scheme may have to be assessed. Bandwidth may be required to request and obtain measurement information from, and to send results to, other locations in the network. In addition, on-line displays associated with a charging scheme may make appreciable demands on computational and bandwidth resources. The use of the charging scheme could risk causing network traffic instability or congestion. With dynamic charging, for example, rich users might pay for the use of the network, whatever the price; this could lead to network congestion. Data given to a charging scheme from the user or from on-line measurements should be secure.

11.1.2.2 Operational Aspects Co-operative Sharing: It is beneficial to the network operator if charging schemes encourage the sharing of network resources by a multiplicity of uses. For example, dynamic charging has a built-in a mechanism for pricing according to the demand for network resources. Other schemes may require the pre-declaration of parameters that can facilitate the effective and optimal management of network resources by the NO to the benefit of the users of that network. For example, parameters such as time of day, week or year can help to encourage customers to have usage patterns that help to reduce the incidence of network congestion. Technical Predictability: The network operator will appreciate a charging scheme that enables it to predict the effect of a particular user on the overall link and network traffic. For example, a band-limited scheme, such as is used for POTS, based on duration charging gives the network operator information about the maximum requirement a user will make on bandwidth resources.

103 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

11.1.3 Financial Implications of Using a Charging Scheme The ultimate purpose of a charging scheme is usually to generate revenue and earn profit for the supplier or operator. A scheme should be responsive to competition and incorporate mechanisms for increasing revenue. A usage sensitive scheme helps to relate costs to resource usage and to build up an appreciation of cost-based pricing. However, a scheme that depends on the content of a call can give the operator flexibility in adjusting prices [D9a][SCH1][SCH2]. The issues of usage sensitivity, profitability in use, peak-load pricing, discounts for high usage and adaptability require an infrastructure that has the capability for accurate and responsive traffic measurements.

11.1.3.1 Issues requiring a Goodsurements Mea Infrastructure Usage Sensitivity: A usage-sensitive charging scheme can help a network operator or service provider to relate the price it charges for a service to the use that the service makes of network resources. Some charging schemes can take into account traffic properties such the peak bit rate and statistical properties such as mean bit rate and burstiness. Factors such as robustness, security and adaptability are relevant to the issue of usage- sensitivity. Robustness: The calculations a charging scheme performs and the results it produces, should be, as far as possible, insensitive to rounding errors, the realisation of algorithms, the realisation of statistical procedures, the programming language used and the nature of the measurements and metering required from the network. Security: As mentioned before, the data presented to the charging scheme should be inherently secure. In addition, it should not be possible to fool the charging algorithm, for example, by taking measurement samples at instances inappropriate to the network operator. Lack of security could also be occasioned if the charging scheme has complex interaction procedures with other network functions such as Connection Admission Control and Usage Parameter Control. Adaptability: Network operators may like a charging scheme that gives them possibility of tuning prices for network resources according to demand. Better still, if the charging scheme can adapt or respond, on-line, to changes in a user's traffic, using an appropriate infrastructure for measuring that traffic. Such a dynamic scheme, is however, usually more complex, and will have additional bandwidth requirement and computational overhead. QoS Sensitivity: Contractual obligations regarding cell loss and delay statistics may require these parameters to be assessed on line, with appropriate price reductions to the customer if these parameters are not kept within bounds. Profitable in Use: The charging scheme should incorporate mechanisms and easily identifiable parameters for increasing revenue. Parameters include, for example, peak load pricing, discounts for high usage, ability to recover incremental costs, price discrimination and ability to response to elasticity in demand. Peak Load Pricing: The operator may wish to price peak loads more heavily than normal loads to try to even out the network traffic Discounts for High Usage: The operator may wish to show an appreciation to high users for their custom. In addition, high users may have proportionally lower administrative and set up costs than smaller users. Incremental Costs: These occur as a result of a new service being introduced. A charging scheme should allow the operator to charge in such a way as to absorb these costs. Price Discrimination and Elasticity of Demand: A supplier or operator may wish to sell the same service to different customers at different prices, even though costs to the operator are essentially the same. This may be

104 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers as a result of customers having different demand elasticities. The operator may increase his profit by charging inelastic users more for a given service. An example of an inelastic user could be a large corporation that wishes to set up a video-conference at short notice and is willing to pay a premium for this facility. Bandwidth Limiting. Schemes (eg. POTS) involving an upper bandwidth limit that is never attained by the customer may be profitable to the network operator. Predictability (Financial: The network operator will usually appreciate a charging scheme that enables it to predict effectively the revenue generated by a customer's use of a particular charging scheme. The network operator can thus have an accurate revenue forecast. At the same time, the network operator has a useful indication as to the values to apply to the appropriate tariff parameters, so as to avoid the risk of customer rejection. Response to Competition: The charging scheme should be flexible enough to allow the operator or service provider to respond to changes in charges introduced by the competition. This process is helped by an appreciation of the extent to which the scheme is usage based and the extent to which it relates to the content provided by the service. Charging by content is based on marketing considerations as opposed to technical aspects.

11.1.4 Systems integration

11.1.4.1 Overview Charging and Billing, in the network context, can be mapped on to the ITU-T Telecommunications Management Network (TMN) architecture [M3010]. There are three processes [D5]: • On-Line Data Capture • Data Modification • Accounting - Charging and Billing

Network Management Platform

Operations System Operations System Network Manager Accounting Server Charging Function Mediation Function Billing Function

Network Element Platform

NE Function Process Data Capture

Figure 11.2: The Charging and Billing Process (Simplified)

105 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

11.1.4.2 Data Capture For charging, data has to be captured, even if it is merely call duration information. In M.3010 [2], Data Capture is a Network Element Function. Data capture takes place in the switching elements. The captured information will be used in a later process to produce the bill for the customer. There are two main types of data: • Contractual Data. This has traditionally been known before inception of a call. • Connection and Usage Data. This is gathered while the call is in progress.

11.1.4.3 Contractual Data As the name implies, this relates to data describing the service contract with the customer. It can comprise the following items. • Installation • Access Rental • Equipment Rental • Time of day charging bands • Call set up charge. • Bandwidth charges in forward and reverse directions. In the past contractual data has not been considered as data that requires capture due to the nature of the existing types of contract. However for the ATM networks of the future, it is envisaged that it will be possible to negotiate the contractual data dynamically on-line, while a call is in progress [D16][SCH2].Contractual data will then have to be gathered on-line in real time. This can have implications for (for example) processor memory and - in the context of data gathering - network bandwidth requirements.

11.1.4.4 Connection and Usage data. This is collected on a per-call basis, and reflects the network resources consumed by the call. Connection and Usage Data can only be collected by the Network Switching Elements.

11.1.4.5 Telephony (POTS) For telephony, the "collection and usage" data is always collected by the switch to which the originator of the call is connected. The parameters are all known to the local switch. They are normally stored in the switch until the end of the billing period at which time they are down loaded to the accounting server for processing. For telephony, the data collected is very simple and consists of: • Calling party identifier • Called party Identifier • Time call commenced • Duration of the call

106 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

11.1.4.6 ATM Network Here there are more parameters to be collected. The parameters are more complex in nature. The data to be collected may include the following: • Identification of the originator and destination of the call • Time the call commenced and terminated. • Type of connection requested and type of connection actually made • For both forward and reverse connections: • Number of cells transmitted, received and discarded. • Peak and mean cell rate • Interruption duration The ‘forward direction’ refers to cell flow from, the ‘reverse direction’ to, the originator of the call. Cells with cell loss priorities 0 and 1 may need to be counted separately. Peak Cell Rate (PCR) can provide measurement challenges. PCR will need to be defined so that it can be measured to give consistent values irrespective of the bandwidth of a connection and irrespective of the way this bandwidth might change. The PCR may only relate to a very small time period of a connection, so that the practical significance of the PCR in the context of network resource being used may need to be assessed. Not all of the data items associated with a call are known to the local switching Network Element. Agreement will be needed about where all the data for a call is to be collected, and what kind of protocols are needed to co-ordinate the data collection process. For TORRENT, the Local Access Point (LAP) may be the focus for data collection, or the LAP may be required to send the data to elsewhere in a distant network - for example to the transmission node of a call originated somewhere in the distant network. Data collection is important in the context of the following situation. With ATM, not all of the cells offered to the network may be transported to the receiving party. Some cells may be rejected at the input or be marked for discard. Hence cell counts are required for transmitted and received traffic. This information has to be co- ordinated across a number of switching elements in order to ascertain the correct information on the number of cells that have been passed. This is so that the customer is not charged for the transport of cells that never reached the destination. In addition, if several operators are involved, the operator who passed the cells successfully should not be penalised, just because another operator was unable to pass the cells on to the destination. Multi-party calls could also present additional complications. Even if multi-party calls can be considered to be a number of associated point to point calls, the accounting server will need to collate individual records for each point-to-point call in order to provide the charge for the whole multi-party call. Detailed information on measuring points in an ATM network is given in [D9a].

11.1.4.7 Mediation Function Once captured, this data may require modification to provide the information required by the charging and billing system. This is done by the mediation function. Indeed, additional parameters needed by charging and billing, may need to be calculated by the mediation function from the raw data collected during the call. The mediation function may also need to re-arrange the parameters into the format required by the billing process. In Figure 11.2, the Network Management platform performs the Mediation Function. This implementation allows for optimisation of the data transfer from the Network Elements, and makes use of the processing power of the Network Manager. The mediation function could also reside in the network element (switch). If

107 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers the mediation device is separate from both the switch and the billing system, it can also be used to interface different manufacturers’ switches to the same billing system or indeed, the same switch to different billing systems.

11.1.4.8 Accounting Accounting comprises charging and billing. The charging algorithm is applied to the data so that the service provider can obtain payment for the services provided. Most network operators are reluctant to discuss the charging process in any detail. Billing makes use of the charging process. Charging and billing are usually performed within a single platform called the Accounting Server.

11.1.4.9 Transfer. Transfer is the mechanism by which information is passed from one process of the system to another and is denoted b the arrows in Figure 11.2. Transfer functions are usually context- and implementation-dependent. Further information on "Transfer" is given in [D16].

11.1.4.10 Classifying Services There are at least three approaches to classifying ATM Services.

• CCITT I.211: Here the International Telecommunications Union considers three basic components: voice, video and data. Each of these is further subdivided into interactive, distributive, conversational, messaging, and retrieval. Since the conversational data service is deemed not to exist., this gives a total of fourteen classes.

• ATMF TN 4.0: The ATM Forum has defined five ATM service categories (ASCs) in terms of bit rates: constant, real and non-real time variable, available and unspecified.

• ITU-T I.371; The International Telecommunications Union has also looked at ATM Transfer Capabilities (ATCs), which are roughly aligned with the ATM Forum service categories. These are again defined in terms of bit rates - the ATM Forum definitions are in brackets where they exist. These bit rates are: deterministic (constant), statistical (non-real time variable), available (available), ATM Block Transfer. As can be seen, there is not a complete alignment between the ATM Forum and ITU definitions. The ITU definitions are simpler than those in CCITT I.211 The user should be able to influence the activation/deactivation of services (incl. firewall settings, …), registration for services, set limits for service specific/overall costs, and see these reflected in the SLA, and the subsequent bill. Charging and accounting has traditionally been an add-on extra for telecommunications services, the main work of the designer being in providing connectivity. Unfortunately this approach has led to a restricted range of options for deriving income which meets the needs of neither the customer nor the provider. The lack of useful charging mechanisms in earlier IP networks has given rise to the "subscription only, flat rate" charging strategy which has proved unacceptable in a wide range of situations; the heavy user of services is undercharged and fails to generate sufficient revenues to meet the network investment required, while the light user is charged at a common rate and feels that the charge considerably exceeds the value he receives. Extensive studies on the needs for telecommunication charging and accounting have been made in the European Community ACTS projects CanCan (Contract Negotiation and Charging in ATM Networks) and Ca$hman. Although these studies addressed mainly ATM charging, the conclusions can be applied to any

108 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers connectionless or virtual circuit environment, including TCP/IP and similar. This, and subsequent contributions to the TORRENT work, will draw heavily on the conclusions from these projects. Before considering the functions which must be incorporated into an access network, it is important to understand what is important both to the customer and the network provider. It might be thought that these two parties have different values, but in a competitive environment the needs converge, as dissatisfied customers may move on leaving no revenue for the service provider. In brief here are the main requirements identified in the CanCan project:

• No surprises: A customer does not wish to receive a bill grossly in excess of the amount his experience would tell him is correct, however much cost is incurred by the network or value obtained by his users. The "undesirability of surprises" applies all the way from the domestic situation to the telecoms manager of a large multi-national company. Charging should not be dependent upon the instantaneous state of network congestion. A user must have some confidence in predicting the cost of the use of a service.

• Price Predictability: The user may wish to estimate the charge before making a call. For example, with POTS, the absolute maximum bandwidth is declared in advance, so that users pay in proportion to be time duration of the connection, plus some set up charge.

• Any given traffic (at a given time of day) should always generate the same charge: Statistical or total traffic based charges may not be satisfactory in this respect, as a user would not wish to be hostage to chance or other users’ activity.

• If the overall telecoms activity remains the same, then this should generate the same charge: Any short term traffic-based scheme which takes no account of customers’ expectations based on historical data, would not meet this requirement. A corollary of this is that the use of a network during congestion can only be discouraged by the poor service offered. There is a potential conflict in the desire of the customer to be charged for what he uses (usage sensitivity), yet at the same time not to have any surprises in the bill (price predictability).

• Ease of Use: The user would like to have a charging scheme that is easy to use and for which the financial implications of using it are in his or her favour.

• User Knowledge Required: How much experience does the user need in order to maximise his benefit when employing the particular service ?

• Traffic Description Needed: How detailed a description (eg. statistical) of the user's traffic is required before the charging scheme is applied? Are measurements of network and traffic parameters needed while the connection is in progress? What onus does this place on the user to supply the information.

• On-line Display: How user friendly is the scheme - in particular, is there some on-line display associated with it?

11.1.4.11 Financial Implications of Using a Charging Scheme

• Profitable in Use: The charging scheme should represent value to the customer in terms of quality of service and price. Hence, experience in using a particular charging scheme could help customers to maximise their financial gains when using that scheme.

109 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

• Usage Sensitivity: Users may appreciate a charging scheme if it is usage-based since this conforms to the traditional notion of paying for the network resources that are actually used. However, it has to be defined what parameters the charging scheme takes into account (eg. peak bit rate and statistical properties such as mean bit rate and burstiness?).

• Dynamic: To what extent, and how quickly, should the charging scheme adapt or respond to changes in a user's traffic? Customers may like a charging scheme that gives them the possibility of tuning and re- negotiating prices for network resources according to demand. They may also like a charging scheme that can adapt, on-line, to changes in a user's traffic.

• Security: A user would like a charging scheme to be secure. For example, a charging scheme that samples bandwidth could allow an unscrupulous operator only to measure traffic when traffic bursts occur.

• Contract: The charging scheme should take account of the contract between customer and provider?

• Auditability: Auditability is concerned with the history of a connection or call. A record of this can be useful in the event of a customer query. It must be possible to audit the basis of a user’s bill. A number of comments on auditability from user perspective are appropriate: Flat rates provide easy auditability but have the disadvantages outlined above. Dynamic charging schemes in which charging is based on the detailed times of arrival of packets provide for difficult auditability as it requires large amounts of data to be recorded. Effective Bandwidth based charging seems to require a complex audit process. Very little discussion seems to have occurred among operators in the past regarding auditability. Auditability is very much a matter of trust. Most users most of the time will accept the bill presented. On rare occasions they will have some doubts and will follow the audit trail. If their concerns are satisfied then that will be the end of the matter. However if it is not possible to satisfy their concerns owing to the absence of the audit trail, then the concerns will fester, paranoia will reign and the user-provider relationship will break up. Thus it is in the providers interest to provide the required transparency, but the cost is an overhead and so charging schemes which are expensive in audit requirements are at a significant disadvantage. In summary, charging schemes should be auditable, practical, usage sensitive and predictable.

11.1.4.12 What is charged for? Traditionally, telecommunications charging has been based on network operators costs, and this has often been reinforced by regulators' dictates. Users are interested in the value they get from the service, although they feel that they have been unjustly treated if they discover that the supplier is making a profit thereby. However there is no converse feeling if a low value service, which is expensive to provide, leads to a network provider's loss! From these considerations there are two basic ways of charging: • Based on the use of network resources - eg. number of bytes, channel capacity, connection time, distance. • Based on the service provided - eg. number of pictures, quality of picture, usefulness of the information. Traditionally the network resource use has been standard but it can be seen that this is becoming of less relevance to users. It is likely to remain a standard technique for wholesale trading between network component providers; thus it is of relevance to the network side of the TORRENT test-bed.

110 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

On the users’ side there will be less need for the monitoring of resources. The scenario is becoming commonplace that a user provides a service provider with a credit or debit card number, either for a single transaction or a continuing series of transactions, and the usage of that service is then charged to that customers’ bank account. The service provider’s need is then for some assurance that the connection is to the person or location to which the agreement applies. Historically, telex machines provided an answer-back code and telephones provide a calling line identity. Nowadays passwords and cookies can provide similar security but some form of network port identification would be a useful further adjunct. Linked to this could be some form of encryption. Even if a user’s mobility requirement means that the process cannot restrict connections to a single port, the identification of ports used in connection with a particular service would enable retrospective investigation of fraudulent use. Against this must be set an individual’s right to privacy. The criteria deemed as most important in one survey (the CANCAN User Forum are: • user knowledge required • predictability • ease of auditing the bill • the fact that the scheme is resource based.

11.2 Support for adding new service provider offerings In order to make the users’ “out of box” experience, as simple as possible, service providers must be able to describe the services they wish to offer. This description must contain at least the following: • Business information, name and address of service provider, contact details (email/web address etc.) and security certificate. • Service information, what a service is, what it does • Technical information, where a service is, what interfaces it supports, its QoS requirements, pricing information etc. By providing this information directories of services and service providers can be built up. This allows new service offerings to be advertised, and for users to search for services based on a number of criteria (location, price, etc.). For example a user may wish to use an online broker in the United States but stream video from a local VoD source. The inclusion of pricing information helps the user make an informed choice, by aiding them to estimate the cost of using a service. In addition to this the suppliers of the RG or network providers may wish to pre-configure certain service and service options. This allows the network providers to give selected service offerings “preferred” status. For users, this is an extra convenience as the service is available immediately. For example an IP telephony provider would obviously pre-configure their telephony service, when supplying the RG.

Service providers could also offer several value added features and services to customers based at the operator’s local exchange. Some of these service-provider-specific offers may include: • Adaptation of presentation / Content reformatting (according to access line capability) done in the local exchange (eg. video compression level adaptation according to available network bandwidth or quality requested by the user; provision of different streams for different bandwidths) • Server based games/applications (SW and processing power available for customer’s rental). • Web hosting

111 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

There are also some improvements that have impact on service quality and that could be integrated at the local exchange. These can be integrated by the network operators or the service providers. Example of these enhancements are: • Firewall Service • Adaptation of presentation / Content reformatting • Server based Games / Applications • Web hosting • Caching (minimise core traffic; improve performance) • Load balancing / Content switching (improve performance and reliability) • Datawarehousing (knowledge about users behaviours) • User store area (user profiles/preferences store) • Redundancy (improve reliability) • Intrusion detection (improve security)

11.2.1 Firewall Services

11.2.1.1 Centralised Firewall Services Centralised firewalls have traditionally been used to protect corporate networks from hostile activity on the Internet. IP Packets to and from the corporate network are inspected to filter out traffic that is not allowed as defined by the company’s security policy. However, a firewall can’t protect the network from attacks that do not go through the firewall eg. an attack from a traitor inside the network. Virus protection is required in addition, as most firewalls do not effectively stop viruses. A centralised type firewall may be placed at the entrance to the home network to filter traffic to/from the home network. A firewall with the above characteristics may also be placed at the local exchange (LAP) in order to filter out suspicious traffic from the Internet. This can be configured with one policy for all users. It is also possible to offer a set of pre-defined firewall services that the users can choose amongst depending on the level of security required thus providing a more distributed type of firewall services. This type of firewall service is available on the market today. It should be noted that this will not provide protection from attacks from a PC within the local community, but will provide the members of the community with protection from “the rest of the world”. Hence, such centralised firewall services can remove a large number of threats and reduce the amount of unwanted traffic to the local community.

11.2.1.2 Personal Firewall Services A personal firewall protects an individual machine and must be administered by the individual. In this case, a firewall software client is downloaded onto the personal computer. The individual must determine the firewall policy and establish the set of rules for which types of traffic are allowed. This is acceptable as long as the individual is knowledgeable about the technology and knows how to set up the rules that prohibit unwanted traffic without causing trouble for the applications that are in normal use. Internet service providers often offer this type of firewall service to the users along with anti-virus protection.

112 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

11.2.1.3 Distributed Firewall Services Distributed firewall services protect individuals and are administered centrally. The firewall policy is downloaded from the central policy manager onto the individual computer. In this case the user does not have to set up the rules as this is taken care of by the central administrator. The policy is centrally defined but enforced on each endpoint. System management software distributes the firewall policy to the individual hosts. Packets are filtered according to the firewall policy and, in the case that the implementation uses IPsec, these packets may be accepted or rejected subject to cryptographic verification of the identity of the sender. Service providers can use distributed firewalls to provide a set of pre-configured firewall schemes. Furthermore, individually tuned firewall services enabling protection from many different points of attack can be supported. This is a valuable service, as the average user may not possess the knowledge to configure a firewall and will appreciate obtaining this type of assistance.

11.2.1.4 Firewall Services in TORRET N It is feasible that the TORRENT testbed can provide centralised firewall services and/or distributed firewall services. Centralised firewall services may be configured at the LAP or the RG. In addition, distributed firewall services may be provided using the LAP as the central administrator. In this case, for security reasons, it is recommended that the central administrator function is installed on a dedicated machine. Such firewall services that can be provided to users by Service Providers and/ or Network Operators should be demonstrated by TORRENT. Firewalls can contribute to better security by reducing the number of security holes, however they can never provide 100 % security and often function as “security blankets”. They can contribute to the security of the network by reducing the risk of attack and getting rid of unwanted/unnecessary and possible harmful traffic. This type of protection should be demonstrated in the TORRENT testbed. Anti-virus protection is also an essential security measure. See also: http://www.interhack.net/pubs/fwfaq/ and http://www.research.att.com/~smb/papers/distfw.html

11.2.2 Adaptation of presentation / Content reformatting Envisaging a scenario where several or different types of access network technologies will be present, it is important to provide mechanisms that perform the adaptation of services and content to network capabilities available. For example, video with different compression levels could be provided to customers in conformity to the bandwidth supplied by a particular access network technology or in accordance with the quality required by the user, what could mean that not all available bandwidth is used. In addition, since a wide range of terminal equipment will be available at customer premises it could be also desirable to implement content reformatting accordingly to the terminal equipment characteristics. Terminals like multimedia phones and web on TV have particular capabilities and special requirements in terms of usability.

11.2.3 Server based Games / Applications Although there are no substantial offers in this area today, there is an increased interest from the service providers on this type of offers. Users would like to have access to games and applications that they can use

113 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers just when they need or desire. Customers are billed on a usage base, which can be time duration dependent. The software, that could be an application or game, is run at the exchange servers. There could be the need to make some state provision at customer’s premises much like Internet cookies.

11.2.4 Web hosting Service providers could offer web site hosting both for private and business use. Offers can go from simple web site hosting to complete e-commerce solutions that include from shopping carts to auto-responders. These services could provide significant cost saves to customers, as service provider resources are shared among several customers.

11.2.5 Caching A local data store could be also provided to cache incoming Internet content. The caching system should be optimised in order to support a large number of users. This caching of popular content provides a faster response time for each user when browsing the Internet and considerably reduces network traffic. Moreover it eliminates redundant content reformatting, described previously. Advanced caching options could also be supported. This enables, for example, the caching of pictures in different transformations to meet the display requirements of different customer equipment. When an HTTP request is received, the type of terminal that requested it is determined and it is delivered with the appropriate file. Some form of local caching at the residential gateway could also be supplied as a mean to decrease the traffic on the access network and improve user experience.

11.2.6 Load balancing / Content switching As a means to maintain high service availability and low response times, techniques like Load balancing and Content switching can be used at the local exchange servers. These are particularly relevant when offering demanding services like e-commerce and Video-on-Demand. Content switching is about centrally controlling at which servers the cache is done and the mechanisms to redirect the requests to the appropriate server locations. Usually associated with it, there is a Load balancing mechanism to distribute the load between the servers. There could be also some content replication at distinct geographical locations for security issues.

11.2.7 Datawarehousing Service providers can use Datawarehousing to obtain general information about their customers’ habits and behaviours. This could be employed to segment their customers according to different usage type patterns, in order to provide a more adequate set of services and offers to specific groups. Moreover, Datawarehousing can provide a efficient way to feed decision support systems (DSS) that can help providers choosing the right strategies.

114 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

11.2.8 User store area A user specific storage area could be provided on the local exchange to maintain information related to the customers, like user profiles. This could include, for example, user quality requirements in terms of bandwidth and costs or configuration parameters of the remote set-up for the environmental home systems. Although this area could be on the local exchange, it could also be supported locally on the residential gateway.

11.2.9 Redundancy To achieve high reliability and stability on the local exchange it is desirable to have redundant equipment, although this leads to higher investments.

11.2.10 Intrusion detection In order to improve security intrusion detection systems can be deployed. These can detect damage and security breaches and sometimes take preventive measures to minimise undesired the effects and prevent proliferation.

115 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

12. Validation Scenarios and Criteria Where feasible and appropriate, the partners will validate that the user requirements mentioned previously in this document can be met by the TORRENT system. The purpose of this section is therefore to identify which services and features could be validated. The precise validation experiments are described in separate Deliverables: D4.2 “Definition of Phase One Field Trials” and D4.4 “Definition of Phase Two Field Trials”. The evaluation results themselves will be documented in the Deliverables: D4.3 “Evaluation of Phase One Field Trials” and D4.5 “Evaluation of Phase Two Field Trials”.

12.1 Examples of Services that could be validated

Access Line NAP Terminal(s) Services Remarks Technology Copper pair Analogue a) Phone + PC Telephony (voice) 8kbit/s voice, the modem b) Screenphone + Internet (e-mail, web browsing, rest for data c) Screenphone + PC IP telephony, FTP, ASP) d) Environmental systems + Environmental ISDN NT1 a) Phone + PC Telephony (voice, multimedia) 0->128kbit/s voice b) Screenphone + Internet (e-mail, web browsing, 0->128kbit/s data c) Screenphone + PC IP telephony, FTP, ASP) d) Environmental systems + Environmental XDSL a) Phone + PC +TV Telephony (voice, multimedia) 8 kbit/s voice, the modem b) Screenphone + PC + + Internet (e-mail, web browsing, rest for data TV IP telephony, FTP, ASP) c) Environmental systems + Entertainment (but not broadcast TV) + Environmental CATV Cable modem a) Phone + PC +TV TV + Internet (e-mail, web b) Screenphone + PC browsing, IP telephony, FTP, +TV ASP) c) Environmental systems + Entertainment + Environmental Powerline Modem a) Phone + PC + CATV Telephony (voice, multimedia) Typically, a data b) Screenphone + PC + + Internet (e-mail, web browsing, interface (Internet CATV IP telephony, FTP, ASP) + IP telephony) is provided to end- c) Environmental systems + Entertainment users. But + Environmental nx64kbit/s (up to 2Mbit/s) is also possible

116 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

Optical Fibre Opto-electric a) Phone + PC + TV Telephony (voice, multimedia) All services are converter b) Screenphone + PC + + Internet (e-mail, web browsing, possible TV IP telephony, FTP, ASP) c) Environmental systems + Entertainment + Environmental

From the above table, we start to see: • what terminals we need, • what interfaces are necessary, • what services we will show, and • which services can be offered over multiple access networks (ie. are interesting for TORRENT).

The inclusion of home network considerations is necessary to complete the picture, though, since these impact (at least) the terminal interfaces, and the services that can be shown.

12.2 Features to be validated A wide range of user requirements were identified in section 4 of Deliverable D1.1. Some of the requirements are more market-oriented (price) and others (eg. reliability, ease of installation, use, control, maintenance, security, response times) are heavily dependent upon the product development phase. These 2 areas (market and product development) are outside of the scope of this project. The TORRENT validation will therefore focus on the novel technical capabilities that are enabled by the system. The key features to be validated are therefore:

• Ability of the terminal equipment to be connected to – and function across - different home network technologies

• Ability of the RG to be connected to – and function across - different home networks and access network technologies

• Ability of the LAP to be connected to – and function across – different access networks and core network technologies

• Ability of the user to select (manually, or through intelligent user preferences) the appropriate access line technology (where a choice is available) for a particular service

• Ability of the system to select the appropriate access network according to the requested instantaneous QoS. By instantaneous is implied that the QoS (and therefore also the access line) can change not only on a call-by-call basis, but also during a session. (This may be user- or network- initiated)

• Ability to adapt (or select) content according to the capabilities of the access network

• Ability to provide system status information to the user (RG) and the network provider (LAP)

• Ability to provide new services fast and easily to customers

• Ability to access home environmental systems remotely - for configuration purposes - with security

• Ability to support user profiling

117 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

13. Conclusions This document has been produced by WP1: Architectural Framework. The initial role of WP1 has been to define the overall framework for the project, in terms of identifying the QoS requirements of typical services, and then mapping these requirements to the network capabilities of a variety of home networks (Home RF, DECT, IEEE 802.11b, Bluetooth) and access networks (CATV, fibre, xDSL, powerline communications, ISDN, wireless optical, LMDS). The overall requirements for service providers, network operators and manufacturers is to have an agreed system architecture that enables the user requirements to be met in an efficient manner. Such an architecture has been described in section 2. This architectural framework is fundamental to the rest of the project. This framework defines the context within which the QoS negotiation and service provision will operate, and it determines how services must be defined so as to enable them to be prioritised on the access network, and later routed onto the most appropriate core network. It is based on the typical network environments that currently exist, but extended with hardware and software functionality to provide more flexibility in the usage of the underlying networks for meeting the instantaneous requirements of existing and emerging services. In order that the architecture can be developed into a fully functioning system, the services (service components) must be specified, and this was done in section 3. The framework defined in this Deliverable will be re-examined throughout the software development process, and especially following the first experiments.

Whilst Deliverable D1.1 listed and analysed the user requirements in some detail, this document has concentrated on how operators and manufacturers can satisfy these user needs through a variety of networking technologies. After the overview of the TORRENT architecture framework, anticipated terminals, underlying networks and new devices (the Residential Gateway and the Local Access Point - introduced in the architecture section) were characterised. The associated protocols, and their capabilities for conveying specific services were then identified and analysed in detail. Each of these items has been documented in an explicit section (sections 4 – 9). The structure of the document led the reader logically through the network from the terminal, the home network, the Residential Gateway, the access network, the Local Access Point, to the core network.

The mapping policies (service components to network resources and technologies to underlying physical infrastructure) were also presented.

Furthermore, this Deliverable introduced the control and management software features that TORRENT will develop to exploit the fact that users will be able to choose from a number of different home-, access- and core- network technologies. This control and management software will enable services to be routed to the most appropriate network, according to the instantaneous QoS requirements of the user. The flexible control and management of services is a key value-added feature of the TORRENT system. However, the architecture is suited for incorporating other features, such as accounting, the support of a firewall at the local exchange, and the easy integration of specific service provider offerings, adapted for the user environment. These were also mentioned in this Deliverable.

The testbed scenarios that will be used to validate the concepts will be defined in future documents. Some ideas of features that would prove the TORRENT concept are presented at the end of this Deliverable, but these should be understood only as possible feasibility trials.

118 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

14. References

[1] URL: http://paul.bristow.free.fr/complify/index.php3?action=overview [2] URL: http://www.3eti.com/3e%20Gatewayfinal.htm [3] URL: http://www.jungo.com/openrg/products.html [4] URL: http://http://www.dlink.com/products/broadband/di701/ [5] OSGi Service Gateway Specification Release 1.0, May 2000. [CAN] CANCAN Consortium, Contract Negotiation and Charging in ATM Networks, ACTS Project AC014, 1995-1998) [D16] CANCAN Consortium, Dynamic Charging Schemes: Network Implications, CANCAN Deliverable D16, AC014/QMW/-/DS/I/370/A7, Document 370in-a7.doc, July 1998 [D5] CANCAN Consortium, ATM Charging Schemes - Review of ATM Charging Schemes and their Performance Issues, CANCAN Deliverable D5, AC014/QMW/-/DS/P/005/b1, Document 312ds-b2.doc, October 1996. [D9A] CANCAN Consortium, Final Report on Static Charging Schemes and Their Performance, CANCAN Deliverable D9a, AC014/QMW/DS/P/350/a7, 1997. [M3010] ITU-T recommendation M.3010 (1996). [SCH1] E. Scharf, First Steps towards the Development and Assessment of Effective ATM Charging Algorithms, IEE Colloquium on Broadband Charging, Ref. No. 96/222, IEE, London, 12th November 1996, pp7/1-10. [SCH2] E. Scharf, Meeting the Challenge of Charging for ATM, British Telecommunications Engineering, Volume 18, Part 2, August 1999.

119 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

15. Abbreviations

ADSL Asymmetric Digital Subscriber Line ASP Application Service Provider ASP Application Service Provisioning ATM Asynchronous Transfer Mode BER Bit Error Rate BRI Basic Rate Interface CATV Cable TV CE Compact Edition CEC Commission of the European Community CLID Calling Line Identification CLID Calling Line Identity CoS Class of Service CPU Central Processing Unit CSCW Computer-Supported Collaborative Work dB decibel DBS Direct Broadcasting Satellite DECT (digital wireless technology) DiffServ Differentiated Service DS Differentiated Service DSCP Differentiated Services Code Point DSLAM Digital Subscriber Line Access Multiplexer DVD Digital Video Disk E-model ERLE Echo Return Loss Enhancement ETR ETSI European Telecommunications Standards Institute EURESCOM FEC Forwarding Equivalent Class fps frames per second FTTC Fibre-to-the-Curb FTTCab Fibre to the Cabinet FTTH Fibre-to-the-Home FYTTB Fibre-to-the-Building GSM Global System for Mobile H/W hardware HAVi™ Home Audio Visual interface HDTV High Definition TV IEEE Institution of Electrical and Electronic Engineers IETF Internet Engineering Task Force IP Internet Protocol IrDA Infrared Device Adapter ISDN Integrated Services Digital Network ISP Integrated Services Provider ITU International Telecommunications Union ITU-T International Telecommunications Union - Telecommunications

120 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers kbit/s KiloBits/second Khz Kilo Herz LAN Local Area Network LAP Local Access Point LCD Liquid Crystal Display LMDS Local Multi-point Distribution Services LSP label switched path MBit/s Mega bits / second MOS MP3 MPEG Motion Picture Experts Group MPLS Multi-Protocol Label Switching ms milli second MUX multiplexer NFAS Non-Facility Associated Signalling NP Network Performance NT-1 Network Termination 1 NT-2 Network Termination 2 NVOD near video on demand OLT Optical Line Terminal (located at central office or cable head-end) ONU Optical Network Units OS Operating System OSGi PC Personal Computer PCI PDA Personal Digital Assistants PHB Per Hop Behaviour PLC Powerline Communication PNNI Private Network Node Interface PON Passive Optical Network POTS Plain Old Telephone Network PRI Primary Rate Interface PSTN Public Switched Telephony Network QoS Quality of Service RG Residential Gateway RSVP Resource Reservation Protocol RTCP RTP Real-Time Transport Protocol SDH Synchronous Digital Hierarchy SIP Session Initiation Protocol SLA Service Level Agreement SME Small to Medium Enterprise SONET SS7 Signalling System 7 STB Set Top Box STM Synchronous Transmission Mode Ta Absolute delay TA Terminal Adapter

121 Deliverable D1.2 IST-2000-25187 TORRENT Requirements for Service Providers, Network Operators, Manufacturers

TCP Transmission Control Protocol TCP/IP Transmission Control Protocol/Internet Protocol TDM Time Division Multiplexing TE1 Terminal Equipment 1 TE2 Terminal Equipment 2 TELR Talker Echo Loudness TORRENT Towards a Realistic End-User Test-Bed Tr Round trip delay TV Television UDP UMTS Universal Mobile Telecommunications System USB Universal Serial Bus VC virtual circuit VCR video recorder VGA Versatile Graphics Adapter VHS VoCable Voice over Cable VOD video on demand VoIP Voice-over-IP VP virtual path WAP Wireless Applications Protocol WDM Wavelength Division Multiplexing WEPL Weighted Echo Path Loss WLL Wireless Local Loop WP Work Package xDSL Generic term for Digital Subscriber Line technology - A/H/S/VDSL

122