Editorial Dru Lavigne, Thomas Kunz, François Lefebvre

Open is the New Closed: How the Mobile Industry uses Open Source to Further Commercial Agendas Andreas Constantinou

Establishing and Engaging an Active Open Source Ecosystem with the BeagleBoard Jason Kridner

Low Cost Cellular Networks with OpenBTS David Burgess

CRC Mobile Broadcasting F/LOSS Projects François Lefebvre

Experiences From the OSSIE Open Source Software Defined Radio Project Carl B. Dietrich, Jeffrey H. Reed, Stephen H. Edwards, Frank E. Kragh

The Open Source Mobile Cloud: Delivering Next-Gen Mobile Apps and Systems Hal Steger

The State of Free Software in Mobile Devices Startups Bradley M. Kuhn

Recent Reports

Upcoming Events March Contribute 2010 March 2010

Editorial Dru Lavigne, Thomas Kunz, and François Lefebvre discuss the 3 editorial theme of Mobile.

Open is the New Closed: How the Mobile Industry uses Open Source to Further Commercial Agendas Andreas Constantinou, Research Director at VisionMobile, PUBLISHER: examines the many forms that governance models can take and 5 The Open Source how they are used in the mobile industry to tightly control the Business Resource is a roadmap and application of open source projects. monthly publication of the Talent First Network. Establishing and Engaging an Active Open Source Ecosystem with Archives are available at the BeagleBoard the website: Jason Kridner, open platforms principal architect at Texas 9 http://www.osbr.ca Instruments Inc., introduces the BeagleBoard open source community. EDITOR: Low Cost Cellular Networks with OpenBTS Dru Lavigne David Burgess, Co-Founder of The OpenBTS Project, describes 14 [email protected] how an open source release may have saved the project.

CRC Mobile Broadcasting F/LOSS Projects ISSN: François Lefebvre, lead for CRC's Mobile Multimedia 1913-6102 Broadcasting team, presents CRC’s attempt to increase 17 collaboration and innovation in the field of mobile broadcasting ADVISORY BOARD: Experiences From the OSSIE Open Source Software Defined Tony Bailetti Radio Project Leslie Hawthorn Carl B. Dietrich, et al, describe a university-based open source Chris Hobbs Software Defined Radio project based on the U.S. Department of 22 Rikki Kite Defense’s Software Communications Architecture. Thomas Kunz Michael Weiss The Open Source Mobile Cloud: Delivering Next-Gen Mobile Apps and Systems Hal Steger, Vice President of Marketing at Funambol, inc., 27 © 2007 - 2010 discusses trends that are driving the adoption of the mobile cloud Talent First Network and the role of open source.

The State of Free Software in Mobile Devices Bradley M. Kuhn, Policy Analyst and Technology Director at the Software Freedom Law Center, discusses the current penetration 32 of F/LOSS in mobile devices.

Recent Reports 37

Upcoming Events 39

Contribute 42 Editorial "Today's modern handset represents a Open source software and hardware has 'melting pot' of communications and mul- become an accepted way of developing timedia technologies." new and interesting applications in many R. Wietfeldt, information and communication techno- logy domains: operating systems, data- Mobile is the editorial theme for this is- bases, Web infrastructure, and sue of the OSBR. This month's authors applications. It's not surprising that with provide an overview of how open source the increasing popularity of mobile hand- fits into the world of handheld mobile held devices, users and researchers have devices, discussing everything from the explored the power of open approaches hardware to the software applications to providing innovative new applications running on the device. Their discussion is and services in this domain. However, un- not limited to mobile phones as they like personal computers and the Internet, cover other aspects of the complete mo- mobile handsets were tightly controlled bile system, including transmitters and re- by mobile network operators (MNOs) ceivers. It is our hope that their insights who developed a vertical ecosystem by in- prompt you to think about open source tegrating the communication infrastruc- the next time you reach for your mobile ture, the handheld device hardware, and device. often the applications installed on those devices. The software and protocols run- As always, we encourage readers to share ning the mobile communications infra- articles of interest with their colleagues, structure and devices are often and to provide their comments either on- standardized by membership-only bod- line or directly to the authors. ies, where large MNOs and manufacturers have a predominant influence. These The editorial theme for the upcoming players invest significant financial re- April issue of the OSBR is Cloud Services sources into shaping the industry along and the guest editor will be Mike Kavis. their vision to gain a competitive advant- Submissions are due by March 20--con- age. A current example is the ongoing tact the Editor if you are interested in a battle about the dominant radio access submission. technology for 4G cellular systems: LTE vs. Wimax. Dru Lavigne These trends have changed recently. Editor-in-Chief Companies such as Google, Nokia, or Openmoko and Industry Alliances such as the are providing the core building blocks, both in hard- ware as well as software, of increasingly Dru Lavigne is a technical writer and IT open mobile devices. This issue of the consultant who has been active with open OSBR reviews the relevant trends in the source communities since the mid-1990s. open mobile platform space from a num- She writes regularly for BSD Magazine and ber of perspectives. As the articles in these is the author of the books BSD Hacks, The issue show, there is a lot of exciting ongo- Best of FreeBSD Basics, and the Definitive ing work that brings the power of open Guide to PC-BSD. source development to the mobile space. This trend is not just confined to the mo- bile devices as there are also efforts in the development of open mobile infrastruc- ture elements and whole systems. 3 Editorial Andreas Constantinou is the Research Director at Bradley M. Kuhn is the Policy Analyst and VisionMobile. His article discusses the importance Technology Director at the Software Freedom of governance models to understand the dynamics Law Center. He briefly reviews the history of of an open source product, constrasting it to the bet- free software in the mobile device space, fo- ter understood role of licences. Using the mobile in- cusing on both software and hardware. A re- dustry as an example, he demonstrates how view of the available alternatives to-date leads governance models can be used by open source him to conclude that users, while able to ac- sponsors to control the development of open source cess open code bases from major companies, products, and argues for more education and clarity are at the mercy of these companies. For a on governance models. number of reasons, true software freedom on mobile devices is, as yet, an elusive goal. Jason Kridner is the open platforms principal archi- tect at Texas Instruments Incorporated. His article Thomas Kunz, François Lefebvre discusses the challenges and successes in establish- Guest Editors ing a vibrant ecosystem around the BeagleBoard, a low-cost, fan-less single-board computer. The ef- Thomas Kunz received a double honours de- forts within this community have allowed the Beagle- gree in Computer Science and Business Admin- Board to become a versatile and powerful open istration and the Dr. Ing. degree in Computer embedded device. Science from the Technical University of Darmstadt. He is currently a Professor in Sys- David Burgess of the OpenBTS Project discusses the tems and Computer Engineering at Carleton project's experiences, which will probably become University. His research interests are primarily the first case of a free software GSM basestation in a in the area of wireless and mobile computing. public cellular network. The article focuses on the The main thrust is to facilitate the develop- challenges of the project, as well as the advantages ment of innovative next-generation mobile ap- of having followed the open source route. plications on resource-constraint, hand-held devices, exploring the required network archi- François Lefebvre leads the Mobile Multimedia tectures, network protocols, and middleware Broadcasting team at Communications Research layers. He authored or co-authored close to 150 Centre, Canada. His article surveys CRC’s attempt to technical papers, received a number of increase collaboration and innovation in the field of awards, and is involved in national and inter- mobile broadcasting by developing and offering national conferences and workshops. Dr. Kunz complete end-to-end free and open source software is a member of ACM and the IEEE Computer toolsets. Society.

Carl B. Dietrich, Jeffrey H. Reed, Stephen H. Ed- François Lefebvre joined the Communications wards and Frank E. Kragh discuss OSSIE, a uni- Research Centre, Canada, in 1999 to lead its versity-based open source Software Defined Radio Mobile Multimedia Broadcasting team. Since project at Virginia Tech. OSSIE software has proven then, he has contributed to numerous national useful for rapid prototyping by industry as well as for and international standardization efforts and published research and education of hundreds of R&D projects. His recent work has focused on graduate and undergraduate students. In addition to creating and developing open software build- examples of OSSIE’s successes, the project’s chal- ing blocks for next-generation mobile broad- lenges and approaches to mitigating and overcom- casting networks, devices and applications. ing them are described. With his team, he launched the CRC mmb- Tools and Openmokast open source software Hal Steger, Vice President of Marketing at Funam- projects. He writes about the future of broad- bol, inc., introduces the cloud computing paradigm casting on his blog Broadcasting 2.0 (http:// as a way to deliver mobile applications and data. His www.broadcasting20.org). Mr. Lefebvre gradu- article discusses trends that are driving the adoption ated from Laval University in Electrical Engin- of the mobile cloud, important components of mo- eering where he also completed his M.A.Sc. in bile cloud infrastructure, and the role of open source. 1989. 4 Andreas Constantinou is the Research Director at VisionMobile. His article discusses the importance of governance models to understand the dynamics of an open source product, constrasting it to the bet- ter understood role of licences. Using the mobile in- dustry as an example, he demonstrates how governance models can be used by open source sponsors to control the development of open source products, and argues for more education and clarity on governance models.

Jason Kridner is the open platforms principal archi- tect at Texas Instruments Incorporated. His article discusses the challenges and successes in establish- ing a vibrant ecosystem around the BeagleBoard, a low-cost, fan-less single-board computer. The ef- forts within this community have allowed the Beagle- Board to become a versatile and powerful open embedded device.

David Burgess of the OpenBTS Project discusses the project's experiences, which will probably become the first case of a free software GSM basestation in a public cellular network. The article focuses on the challenges of the project, as well as the advantages of having followed the open source route.

François Lefebvre leads the Mobile Multimedia Broadcasting team at Communications Research Centre, Canada. His article surveys CRC’s attempt to increase collaboration and innovation in the field of mobile broadcasting by developing and offering complete end-to-end free and open source software toolsets.

Carl B. Dietrich, Jeffrey H. Reed, Stephen H. Ed- wards and Frank E. Kragh discuss OSSIE, a uni- versity-based open source Software Defined Radio project at Virginia Tech. OSSIE software has proven useful for rapid prototyping by industry as well as for published research and education of hundreds of graduate and undergraduate students. In addition to examples of OSSIE’s successes, the project’s chal- lenges and approaches to mitigating and overcom- ing them are described.

Hal Steger, Vice President of Marketing at Funam- bol, inc., introduces the cloud computing paradigm as a way to deliver mobile applications and data. His article discusses trends that are driving the adoption of the mobile cloud, important components of mo- bile cloud infrastructure, and the role of open source.

How the Mobile Industry uses Open Source

"Open source licenses tell only half the to access, modify, distribute and contrib- story. The governance model, the implicit ute to the software (http://gnu.org/philo rules defining transparency and influence sophy/free-sw.html). These freedoms into an open source project, is the small have been expanded into ten cardinal print that determines the power dynamics points that form the criteria that every around that project." open source license must adhere to, and Andreas Constantinou which are defined by the Open Source Initiative (http://opensource.org/docs/ Openness is a much-misunderstood osd). word. It represents a kind of good-will moniker to which people attach an im- A handful of different licenses (http://op pressive variety of definitions: open ensource.org/licenses/alphabetical) are source code, open standards, open hand- used in the vast majority of open source sets, openness as in transparency, shared projects; namely the GPL, LGPL, APL, roadmaps, open application program- EPL, MPL, BSD and MIT. Interestingly, ming interfaces (APIs), open route to mar- the GPL which is known as a strong co- ket, and so on. It is a very forgiving term pyleft license (http://en.wikipedia.org/ as far as definitions go. wiki/Copyleft), is most often used in per- sonal computer and Internet projects, One of the mobile industry’s favourite fa- but is rarely used in the mobile industry. cets of openness is open source code. Instead, the licenses used most often in Since 2007, tens of mobile industry giants mobile software are weak copyleft, such and consortia have embraced open as the EPL, or permissive licenses, such source in some form or other: the Symbi- as the APL, due to handset manufacturer an Foundation (http://symbian.org), concerns for downstream liabilities. LiMo Foundation (http://limofoundation. org), Google’s Android (http://android. Licenses are Half the Story com), Nokia’s Qt (http://qt.nokia.com), Apple’s WebKit (http://webkit.org) and What’s often missed in open source dis- Nokia+’s Meego (http://meego.com) cussions is how open source licenses tell are the initiatives that have hit the in- only half the story. The governance mod- dustry front pages. On the surface, these el, the implicit rules defining transpar- initiatives use open source licenses, but ency and influence into an open source that only tells half the story. Behind the project, is the small print that determines scenes, Google, Apple, Nokia and others the power dynamics around that project. use restrictive governance models and control points that effectively detract the In practice, mobile open source initiat- very freedoms that open source licensing ives use a variety of control points - such is meant to bestow. This article discusses as trademarks, private lines, distribution the many forms that governance models of derivatives, ownership of reviewers, can take, and how they are used in the gravity of contributions and contributor mobile industry to tightly control the agreements - to turn an egalitarian gov- roadmap and application of open source ernance model into an authoritarian one. projects. Such control points can detract the very freedoms that open source licenses are What is Open Source? meant to bestow, examples of which are shown in Table 1. Open source licensed software carries four basic freedoms that provide the right 5 How the Mobile Industry uses Open Source

Table 1: Control Points Used in "Open" Mobile Projects

In general, open source licenses pertain The picture that emerges is one where: to the use of the source code in a purist sense, while governance models relate to • open source licenses (the large print the use of a product in a practical sense. that covers source control) are widely Governance models are much more fun- used, converged and well understood, damental in determining the ecosystem while dynamics around that product. • governance models (the small print Mapping Licenses vs. Governance Models that governs product control) are pro- prietary, diverging and poorly under- Governance models can be simplified to stood indicate the democracy of transparency and influence on an open source product. The Maze of Governance Models On one end of the spectrum are autonom- ous trust communities where opinion There have been many attempts at classi- leaders influence the direction of the fying governance models, but there is product through a chain of trusted review- really no universal dictionary, no certific- ers. An example of this type of model is ation body, and an excessive amount of . On the other end are single-spon- open marketing hype to help obscure sor communities where the product rather than enlighten the mobile in- roadmap, private lines, contributions and dustry. This is further exacerbated by the trademarks are controlled by a single fact that governance model particulars company. Based on this simplified gov- are usually available under confidential- ernance model, we map the most popular ity terms. They are rarely critisized and mobile open source projects, as seen in debated in public, in the way, for ex- Figure 1. 6 ample, an intrusive privacy policy would be. How the Mobile Industry uses Open Source

Figure 1: License vs Governance Models

How do open source initiatives control ac- • How is the roadmap formed and who cess to a product, manage influence has voting rights? mechanisms, or intellectual property (IP) rights for contributions? We have attemp- • Is there a transparency of contributions ted to provide some of the key questions in terms of the level and type of contri- that should be asked when determining butions by partipating companies? the level of openness of a governance model: • Is the amount of contributors and re- viewers/maintainers balanced among • Is the latest source code publically avail- the community or does one company able or are feature-packed private code form the gravity centre of contribu- lines available discriminately to selected tions? parties? • Can anyone become a code reviewer or • Is the product roadmap publically avail- maintainer? (upstream influence) able or to members only? • Can contributions be deemed as man- 7 datory? (downstream influence) How the Mobile Industry uses Open Source

• Are derivatives available widely, or through a proprietary route to market, such as seen in the Android Market?

• Can you fork the software and buy the service from somewhere else?

• Is there a trademark applying to the use of the project name and embodiments of products? Andreas Constantinou is the Research Dir- ector at VisionMobile where he oversees • Do contributions require copyright the company's research, advisory and in- assignment or patent grants? dustry mapping projects. He has ten years experience in research, development and • Are there any safe harbour provisions strategy in mobile, specialising in the (http://en.wikipedia.org/wiki/Safe_ handset ecosystem, software strategy and harbor) for contributors to the source open source. Andreas has worked on sever- code? al product and marketing strategy pro- jects for clients including Sony Ericsson, We believe that governance models are RIM, Microsoft, France Telecom, T-Mo- the key determinant of the ecosystem dy- bile, OMTP, Qualcomm, ST Ericsson, namics, and the level of innovation Gemalto and Trolltech and authored nu- around open source projects. merous research reports for analyst firms Informa, Ovum and ARCchart. Andreas A Call for Education also teaches the Mobile Open Source work- shop, part of VisionMobile’s 360 degree What the mobile industry needs is not training courses on complex industry sec- more marketing hype around the benefits tors. Prior to founding VisionMobile, of openness, but more education and clar- Andreas spent 3 years at Orange’s Re- ity on governance models. The industry search and Innovation division, including also needs a benchmark – a sort of open- serving as a technology lead for the Or- ness index - for determining the true dy- ange-Microsoft relationship. His interests namics of an open source product, and include uncovering under-the-radar in- for pushing the corporate sponsors to dustry trends and pursuing human-cent- play fair. At VisionMobile (http://www. ric design. When not hopping on planes, visionmobile.com) we have been quietly Andreas spends his time in Athens, Greece. working towards developing an openness Andreas holds a Ph.D. in Image & Video index and are keen to hear from compan- Compression from the University of Bris- ies who want to make this happen. tol, UK.

Governance is one of the most under- stated topics in the open mobile industry today, yet one of the most fundamental in the direction the industry will be taking.

8 Engaging an active Open Source Ecosystem

"If you’re a developer, you want to be However, I am certainly surprised by where you can fulfill your vision. If you’re what people do with their BeagleBoards. a consumer, you want to be where you can Some examples include: do what you want to do. If you’re a hand- set manufacturer, you need to be where • an Iron Man costume with digitally con- that innovation happens." trolled air-rocket launchers Troed Sångberg http://blogs.sonyericsson.com/ • amateur radios that analyze wave con- troedsangberg/speed-of-innovation tents using software

There should be little doubt that any giv- • robots that know how to stand and walk en area of technology will eventually be on their own occupied by open solutions. Rather, it is a question of “who” and “when.” In mobile • three dimensional digital cameras and embedded platforms, there is not a single dominant proprietary solution to • autonomous flying vehicles that recog- displace, unlike desktop software. In- nize objects around them to avoid colli- stead, open software produces interface sion and redirect their path and compatibility experiences on par with proprietary software. • miniature presentation projection sys- tems that utilize 3D objects and anima- At points where no proprietary solution is tion to keep the audience interested clearly differentiated from open solutions and the barriers to participation are kept • high-definition (HD) media centers low, open innovation thrives at the fore- front of the technology. This article intro- • car computers with touch-screens duces BeagleBoard.org, a project that showing video and virtual instruments creates powerful, open, and embedded with data logging devices based on the BeagleBoard hard- ware, a low-cost, fan-less single-board • touch-screen computers that hang on a computer. By lowering the barriers to par- refrigerator ticipation and making commitments to support and sustain the architecture to This is just a small pool of the hundreds preserve the innovation from that parti- of ongoing projects driven by thousands cipation, BeagleBoard.org has built an act- of hobbyist and engineering individuals ive and growing community of hobbyists and team members. People want to do and professionals advancing the state of more with their mobile or low-power em- the art in low-power embedded comput- bedded computing platforms than place ing. phone calls. They want to collect and pro- cess data from the physical environment What do People do with a BeagleBoard? and manipulate that environment. They want to move data over a variety of me- I am asked frequently if I am surprised by dia and in a variety of formats. the rapid growth of the BeagleBoard.org community. What does not surprise me is When given a computing platform that is the number of people who quickly recog- sufficiently high-performance, low- nize that there is something different power, expandable, affordable, small, about the BeagleBoard – it is something and supported by tools and software accessible and capable – and then want to stacks, developers with the capability to get involved. 9 Engaging an active Open Source Ecosystem extend that computing platform, either The Cheap: are characterized by the de- through hardware or software, will simply sire for access to be universal or at least do so. BeagleBoard.org creates an open have access be of minimal cost to them. innovation ecosystem that advances everyone using the OMAP™ 3 processor- Freedom Crusaders: are characterized based BeagleBoard by encouraging open by seeking prevention of any control over source developers to utilize all the avail- them. able tools wherever it makes sense in their designs. We discuss each of these participants fur- ther below and describe how they benefit Understanding the Community from the BeagleBoard ecosystem. Community Participants A key benefit to collaborating with open source developers is that they are vocal To meet the demands of Community Par- about their specific wants and needs. One ticipants, it has to be clear from the be- of the best ways to attract these de- ginning that the community is large velopers is to simply listen and meet their enough to justify their time. Given that demands. the BeagleBoard is based on the OMAP 3 platform (http://en.wikipedia.org/wiki/ LugRadio (http://lugradio.org), which Texas_Instruments_OMAP), the first was a bi-weekly audio download series broadly available device with an ARM® that featured the com- Cortex™-A8 CPU, little convincing was munity manager (http://www. required to reflect that there would be a jonobacon.org), nicely summarized the large pond of participants. primary characteristics and motivations of open source developers. Each of these That same processor will be used as the characteristics is a way in which the de- basis for the next generation .org veloper or innovator is able to gain some device. This signaled to the Maemo open aspect of control or influence. The types source developer community that they of open source developers can be sum- would get early access with the confid- marized as: ence that their time was being spent in an area where plenty of other developers Community Participants: are character- would notice their work. An English-lan- ized by enjoyment of attention and being guage mailing list was created to be a fo- part of something big, where they can rub cal source of announcements and elbows with influential people and im- queries about community developments. prove their own profile within their de- A single RSS feed was created to aggreg- sired sphere. ate the news published on blogs of de- veloper activities. Today, there are Tinkerers: are characterized by a desire to thousands of mailing list subscribers and know how something works and have thousands of articles about BeagleBoard enough control to change how it works or activities with many answering complex how it is assembled. questions about how to design something with the BeagleBoard. The Underdog Fans: are characterized by a RSS feed provides a centralized location desire to see the dominant influences in a to view the latest articles, and several realm of technology challenged by up- new articles are published every day. and-comers that limit the control of any single entity. 10 Engaging an active Open Source Ecosystem

The Tinkerers Underdog Fans

Many are tired and frustrated with the Underdog Fans are attracted to the chal- way computers work, or don’t work. lenge that BeagleBoard represents to the Tinkerers were born to do something entrenched virtual monopoly held by the about it. They love source code, but processors used in typical desktop com- source hardware is even better. The puting. As embedded microprocessors source code to the operating system and utilizing multi-vendor ARM (http://en.wi the source hardware design files describe kipedia.org/wiki/ARM_architecture) pro- how the BeagleBoard operates and how it cessor cores move up in performance at is built. The source allows Tinkerers to much lower power points, lower prices, not only understand the inner workings, more integration and specialized pro- but to alter them. Dozens of systems that cessing resources for some markets, are open for adding more software or many designs that would have used a tra- hardware have been built using the ditional PC architecture are now choos- BeagleBoard as a starting point, with each ing to switch. Fans help notify those who adding some value beyond the Beagle- have not considered such a switch that a Board while growing the shared software viable solution exists. With the influence base that enhances what the BeagleBoard of these Fans, who are frequently key in- can do. The BeagleBoard is based on a fluencers in medium and large organiza- processor with multiple vendors, which tions and developer communities, the further extends the sources that can be BeagleBoard is being picked up by used by Tinkerers to build and experi- people migrating from: i) desktop or serv- ment with software. Today, there are sev- er operating systems like Ubuntu, Debi- eral OMAP 3 processor-based platforms an, and Gentoo; ii) boot firmware; iii) built from the BeagleBoard with the flexib- media center software; and iv) many oth- ility to add new capabilities, such as those er components not traditionally targeted seen in Table 1: at embedded processors.

Developers have the ability to add many The Cheap other physical world interfaces, all in a variety of form-factors, and many much The needs of the Cheap are partially satis- smaller than the BeagleBoard itself. The fied by the bring-your-own peripherals BugLabs Bug 2.0 (http://buglabs.net) approach taken by the BeagleBoard. In- provides a plugin module architecture stead of being constrained to a single with an online store for people to sell configuration where everyone needs to their own modules that utilizes OMAP 3 pay for all the possible features of the sys- and BeagleBoard support to run the An- tem, the BeagleBoard makes use of off- droid operating system. the-shelf expansion via standard inter-

Table 1: Possible BeagleBoard Capabilities

11 Engaging an active Open Source Ecosystem faces, allowing developers to utilize com- • MontaVista’s (http://www.mvista.com) ponents that they already own or allowing MVL6 and Montabello them to choose the components they prefer. There has been some resulting • Timesys’s LinuxLink (https://linuxlink. complexity for users, yet the lower-cost .com) generates a far greater reach and lower threshold for purchasing decisions. The • ARM’s (http://www.arm.com) ARM limitations created also open doors for Linux Internet Platform and D5 value-added resellers, such as Special Computing (http://specialcomp.com), • Halcon machine vision from MVTec who become participants in the com- Software GmbH (http://mvtec.com) munity. Although Special Computing has been successful creating some bundled • MPC Data’s (http://mpc-data.co.uk) kits with the BeagleBoard, the overall suc- Windows® Embedded CE cess and adoption of preconfigured kits has been limited. The wide variety of de- • QNX (http://www.qnx.com) mands could be to blame, and improve- ments to the BeagleBoard.org website in • Ingenient’s (http://ingenient.com) 720p positioning bundles could be a resolution. TMS320C64x+ processor-based video codecs in Android What has taken hold is the use of the BeagleBoard in a large number of demon- • ARM-based video codecs from Visual strations for various hardware or software On (http://www.visualon.com) products. Some examples include demon- strations of hardware products interfaced • Softkinetic-Optrima (http://softkinetic- to the BeagleBoard: optrima.com) OptriCam

• Hillcrest Labs Freespace (http://hillcrest • ASTC’s (http://www.astc.org) Beagle labs.com/products/freespace.php) in- Board simulator air pointing devices • RidgeRun (http://www.ridgerun.com) • TI DLP Pico Projector development kit (http://focus.ti.com/dlpdmd/docs/dlp The Freedom Crusader discovery.tsp?sectionId=60&tabId=2235) One challenge that must be met to keep • TI eZ430-Chronos wireless watch devel- the Freedom Crusader happy is to ensure opment tool (http://focus.ti.com/docs/ BeagleBoard is not locked down to a toolsw/folders/print/ez430-chronos. single vendor. A key difference between html) with many personal and environ- the BeagleBoard and other ARM Cortex- mental sensors A8 development platforms today is that all of the components are available in low Other examples include demonstrations quantities through the catalogs of elec- of software products or concept demon- tronics distributors, and the documenta- strations: tion for the critical components is available for free online. Also, since ARM • Android implementations from more processors can be licensed by just about vendors than I can count, including the any silicon manufacturer or vendor, it is ARowboat.org project conceivable to even replace the core pro- cessor. 12 Engaging an active Open Source Ecosystem

There are programmable elements avail- The readily apparent innovations in the able to the ARM processor for which there available commercial and/or open are programming guides and freely avail- source products look great, but how do able tools. Thanks to the open program- we understand the full value of com- mability of the digital signal processor, munity involvement? Perhaps the best in- which is optimal for processing video and dicators are the customers who have audio data, projects such as the open already seen the BeagleBoard and likely source Ogg Theora (http://.theora.org) played with it before the first sales call is video decoder thrive and provide the Cru- ever made for a new design. saders with something to crusade about.

Building a Community Around Active Participation

The great success of building a com- munity on the BeagleBoard comes not only from meeting the needs of the open source and open hardware developer communities, but by having core de- velopers act as key participants in both the culture of quality products and the culture of open. Whenever tough de- Jason Kridner is the open platforms prin- cisions need to be made, the default cipal architect at Texas Instruments Incor- choice is the one that involves getting out porated. He is passionate about pervasive of the way of innovators and letting the and accessible computing platforms. Krid- community lead. The more an individual ner graduated from Texas A&M with a participates and contributes, the more bachelor’s degree in Electrical Engineering ownership is required on the direction of and was drawn by the allure of digital sig- the project. nal processing to TI in 1992. He began as a hardware developer, working on board, Further, continued improvements and in- FPGA, and ASIC designs. Utilizing soft- novation are required to keep people ex- ware experience prior to TI, Kridner cited. This summer, the next revision, the transitioned to lead software development BeagleBoard XM, will be running at 1GHz of low-power media software, audio pro- or higher with enough RAM to natively re- cessing, file systems, USB drivers, digital build its entire application stack. This is rights management, and video codecs. He an important milestone in the evolution now defines software architectures that of any computing hardware. The choices enable a broad body of developers on TI's made in the design of the BeagleBoard ARM and DSP based catalog processors. XM reflect the guidance of the com- munity members. Despite having some features that have an effect on the cost for the new product, the commitment to maintain the existing product at the exist- ing price point is being held. This sort of commitment to the existing community is absolutely required.

13 Low Cost Cellular Networks with OpenbTS

"What would the world look like if cellular story for Glenn, and asked how we networks were open-sourced?" should start our search for funding. After Alec Saunders listening and considering, he said, "You http://saunderslog.com/2009/09/01/op don't need money at this point. You need ensource-meets-gsm-at-the-burning-man a movement. Do this as a public open source project and develop a following, then the money will be much easier." In mid-2007, Kestrel Signal Processing, Glenn also suggested using the Burning Inc., a small software radio consulting Man festival (http://burningman.com) as shop in northern California, started writ- a technical proving ground for the pro- ing an implementation of a GSM (http:// ject, for the exposure that it would pro- en.wikipedia.org/wiki/Gsm) basestation. duce, and because it would be a lot of The initial developers were Kestrel co- fun. This one meeting set the direction of founder Harvind Samra and myself. Our our activities for the next 18 months. goal was to create a new kind of light- weight cellular network that could be Despite the name, OpenBTS is not a nat- built out inexpensively in remote and ural candidate for a public open source sparsely populated areas. Our software-ra- project. For such a project to have mass dio GSM system, now called OpenBTS appeal, it needs to be something that is (http://openbts.sourceforge.net), was re- easy to use and can provide significant leased publicly under the GPLv3 license functionality early in the development in September 2008 and will be used in pi- process. A GSM network stack lacks both lot deployments with small operators by of these features. First, the stack must be the time this article goes to publication. mostly complete in layers 1 and 2 (http:// This will probably be the first use of a free en.wikipedia.org/wiki/Osi_model) to do software basestation in a public cellular anything useful and these are the most network, where both network operators complex layers of the system. Second, a and subscribers can download and read developer needs about US$1,300 worth the full source code of the GSM protocol of radio equipment, a legal way to oper- stack that connects their handsets to the ate that equipment, some test phones, rest of the world and where the operators and a working knowledge of how GSM will be free to modify the system to meet networks and handsets interact. Granted, their specific needs. This article intro- US$1,300 is a remarkably small sum for a duces the goals and evolution of the cellular basestation development kit, but OpenBTS project. this cost would be a significant barrier for many potential contributors. Given these History circumstances, we knew that we would not make a public announcement of our To understand the relationship between project until layers 1 and 2 were com- OpenBTS and open source development, plete. We assumed that open source par- you must first understand the project's ticipation would be limited to people history. Initially, we planned to fund the who already owned the required hard- development of low-cost cellular techno- ware, the Universal Software Radio Peri- logy through the standard Silicon Valley pheral (USRP, http://en.wikipedia.org/wi startup process. Our process started in Ju- ki/USRP), for other projects. We would ly 2007 with a meeting of the project have to write a lot of OpenBTS ourselves, founders and our first senior advisor, see it work and then release it, and that Glenn Edens, a former executive at both would take several more months. AT&T and Sun. We had breakfast along the San Francisco waterfront, laid out our 14 Low Cost Cellular Networks with OpenbTS

Even after that, we did not expect a lot of News of the project spread quickly in the outside contributions. USRP and GNU Radio communities and within a few weeks John Gilmore, one of In May 2008, just as the software was be- the founders of the Free Software Found- coming complete enough for a public re- ation (FSF, http://fsf.org) and a backer of lease, a former consulting client sued GNU Radio, expressed interest in the pro- Kestrel and me personally over the Open- ject. John suggested that we transfer the BTS project, claiming that the yet-unre- copyrights for the OpenBTS source code leased source code was misappropriated to the FSF for release under the GPL li- from work I had done for him in the peri- cense. We soon realized that the FSF's od 2005-2007. The work in question was a standard assignment contract preserved GSM stack for an IMSI-catcher, a device nearly all of our rights as developers used to perform false-basestation and through the grant-back of a blanket li- man-in-the-middle attacks on cellular cense. We formalized our FSF copyright networks. The client filed this suit having assignment on October 24, 2008 and seen only a small fraction of the OpenBTS made a GPL release through the GNU Ra- source code and knowing that I had writ- dio web site within a few days. ten other GSM protocol implementations for other defense-sector clients prior to In early November 2008, after tens of working for him. Among other things, he thousands of visitors to the OpenBTS accused me of violating the trade secrets web site and hundreds of downloads un- act, even though GSM is a publicly der GPLv3, the ex-client and his lawyers available specification and IMSI-catch- discovered the public distribution of ers are the subject of patents in the UK OpenBTS. They immediately petitioned and EU. This client eventually claimed the court for an injunction blocking dis- the text of the source code itself as a trade tribution of the software, not against the secret, even though we did not use that FSF, but against the project founders. client's source code in the OpenBTS pro- The court granted a more limited injunc- ject. It was our position that the case had tion requiring us to preserve the "names no merit. We laid off our two junior em- and internet addresses" of those who re- ployees to free up money for legal fees ceived technical information from the and continued moving forward as well as project. While that compromise seemed we could. reasonable to the court in principle, the practical reality was confusing. We could In September 2008, we ran a GSM test net- only distribute the source code to indi- work at the Burning Man festival in viduals through e-mail, but the FSF, not Nevada. Our funds and equipment were being a party to the case, continued to limited, but the system actually worked. run their anonymous access servers un- We immediately followed the test with a affected. We could continue to develop public announcement and the first public privately, but could not contribute to the release of the source code. We made our public distribution of the project we had initial announcement in the GNU Radio started. While this arrangement distorted project's mailing list (http://www.gnu the project's management, it was clear radio.org), a large forum for USRP de- that the only way anyone could really velopers. shut down the OpenBTS project would be to sue the FSF directly.

15 Low Cost Cellular Networks with OpenbTS

Another effect of the injunction was great- that work. er public interest in the project. News of the injunction became a starting point for Public Image many online discussions of intellectual property law. As the public profile of the One of the best reasons for a public re- lawsuit rose, the client was becoming a lease is to maintain the public visibility of well-known member of an industry that the project. By making the source code avoids publicity. The case was settled and available to developers and students for the injunction lifted in August 2009, after inspection and evaluation, we make the several months of negotiations. While the project real and exciting in the minds of a settlement and negotiations were confid- lot people. Since our long term goal is to ential, we have no doubt that our public provide cheap communications in poor release through the FSF had a tremend- and remote areas, it is counter to the spir- ous influence on the early settlement of it of the project to tell students and devel- the case. opment workers that they must pay thousands of dollars to use the software OpenBTS Today in their experiments. From a business standpoint, there is little to be gained by OpenBTS continues to make GPLv3 re- denying groups that can not afford large leases through the FSF GNU Radio pro- licensing fees access to the software. The ject. To date, the public and commercial direct business advantage is that their releases of OpenBTS are identical in func- work may lead to new applications and tionality, differing only in licensing. Our markets that we would not have been long-range plan is to follow a model simil- aware of otherwise. There is also a busi- ar to that of Digium (http://digium.com) ness advantage in creating a public im- and Asterisk (http://www.asterisk.org), age of the project as supportive of groups with a free public release under the GPL with socially important goals. and a commercial private release under a different license. We will also have a con- Testing and Documentation tributor's license agreement that protects the rights of contributing developers While we do not get a lot of no-cost con- while transferring the copyrights for their tributions from outside developers, we contributions to Kestrel. We do not do do get a lot of useful bug reports and de- this expecting other developers to write bugging discussions. By following the typ- our system for free. A great majority of the ical sequence of questions from a new current code has been written by Kestrel user and reading the material posted to employees and paid contract developers the public wiki and discussion lists, we and we do not expect that to change. develop better documentation. The feed- There are, however, several practical reas- back we get through the public release is ons for maintaining a free public release. valuable in development and in the pre- paration of training materials and docu- Free as in Freedom mentation.

Free software and open source software Recruiting are not the same thing, although Open- BTS's public release is both. One of the As our project attracts more funding and great lessons of our lawsuit experience is we are required to hire staff, we expect that widespread public distribution is an the public work on the project to be a use- excellent way to protect your right to con- ful recruiting tool. tinue producing that work. 16 CrC Mobile broadcasting F/LOSS projects

While we do not depend on outside de- "By His Genius Distant Lands Converse velopers to move the project forward, And Men Sail Unafraid Unto The Deep." such developers do exist. Most would con- Epitaph on the memorial tablet tribute more if they could be paid to do so of the grave of Canadian radio and some would be glad to have full-time pioneer Reginald Fessenden jobs on the project. From a hiring stand- point, these developers are known quant- The Communications Research Centre ities. We have been coordinating with Canada (CRC, http://www.crc.gc.ca), the them and already have samples of their federal government’s primary laboratory work. We have had good results hiring for advanced telecommunications re- contract developers this way and hope to search and development (R&D), has been continue and expand that practice in the at the forefront of new developments in future. mobile digital broadcasting technologies since their inception in the late 1980s. Summary Comments During this time, digital replacement technologies have been standardized in Simply put, an open source release may an effort to rationalize spectrum use and have saved our project. Now that the litig- enhance broadcasting applications with ation threat is over, open source work datacasting services and associated pro- continues to serve our commercial in- gram information. Eureka DAB (http://en terests. .wikipedia.org/wiki/Digital_Audio_Broa dcasting) was the first all-digital mobile broadcasting technology to be conceived, developed and deployed. It became a widely adopted standard for digital radio in many countries around the world in the mid-1990s. DAB was officially launched in Canada in 2000.

This paper presents CRC’s attempt to in- David Burgess is a Partner at Kestrel Sig- crease collaboration and innovation in nal Processing, Inc. and Co-Founder of the field of mobile broadcasting by devel- The OpenBTS Project. David has nearly 15 oping and offering complete end-to-end years of experience in signal processing sys- Free, Libre and Open Source Software tem development and scientific comput- (F/LOSS) toolsets for the transmission ing. Much of his work in recent years has and reception of DAB and FM/RDS (http: been in the areas of signals intelligence, ra- //en.wikipedia.org/wiki/Radio_Data_Sys diolocation, and navigation. He has also tem) applications and services. worked in electronic warfare, image pro- cessing, high-fidelity audio processing, Background and DSP system design. He holds an M.S. degree in computer science, and a B.S. de- Radio has always been mobile. AM and gree in electrical engineering, both from FM receivers were introduced in cars in the Georgia Institute of Technology. the 1930s, and later, with the emergence of transistor receivers, into portable devices. Radio was one of the first wire- less telecommunication applications available to the masses.

17 CrC Mobile broadcasting F/LOSS projects

Since then, there has been a growing in- When we created non-standard applica- terest in mobile broadcasting services. tions, there was no way to signal their This has resulted in the proliferation of presence in the multiplex. Application new broadcaster-led standards as well as signalling was hard-coded in commercial standards developed by the mobile in- multiplexers. This was the motivation to dustry. These industries recognize that develop a more R&D and innovation physical layer broadcasting (as opposed friendly multiplexer. Over the years, this to pseudo-broadcasting like multicast on project evolved to become a full-blown the Internet) is efficient, both spectrally F/LOSS DAB multiplexer, now known as and in terms of infrastructure, for delivery CRC-DABMUX. that targets large audiences. This is partic- ularly attractive for media-rich content Another key component of the DAB trans- like radio and TV, which are very expens- mission chain is the modulator. Again, ive to deliver through 2.5 or 3G infrastruc- these types of devices used to be quite ex- tures. To mobile network operators pensive. However, with the emergence of (MNOs), this translates into lower deliv- software defined radio and accessible ery costs and relief on their one-to-one platforms, the implementation of a DAB networks, which remain dedicated to software modulator has become straight- high-margin personal communication forward. The USRP (http://en.wikipedia. services. org/wiki/Universal_Software_Radio_Peri pheral) and the GNU Radio open source For broadcasters, mobile broadcasting is framework (http://gnuradio.org/redmine a natural extension of their traditional /wiki/gnuradio) were invaluable F/LOSS mandate and expertise: cultural content, tools that permitted the integration of radio, public services like weather fore- our CRC-DABMOD modulator in a very casts, traffic conditions, emergency in- short time frame. formation and so on. New opportunities like mobile TV and datacasting are also of Together, these two components can great interest. now generate DAB-compliant signals at an unprecedented accessibility level. We Open Eureka DAB Transmitter pushed this one step further by putting most of our DAB transmission tools on Broadcast transmission equipment is ex- an Ubuntu-based live CD which we dis- pensive due to low sales volumes and rel- tribute for free from our projects' Web atively high technical complexity. This sites. The CRC mmbTools Live CD (http:/ creates a high barrier to entry that limits /mmbtools.crc.ca/content/view/30/54/) the potential for innovation in broadcast- can be launched without prior installa- ing. tion on the host PC to produce DAB sig- nals in real time with a simple mouse At CRC, we realized that application in- click and a connected USRP. novation for Eureka DAB was difficult when using typical commercial transmit- We also used our software to provide ter equipment. The earliest issues ap- DAB audio encoding and multiplexing peared with the service multiplexer. In functionality over the Web. These tools DAB, multiple services are combined in replicate typical functionality of physical real time to form a single bitstream or devices and are known as Web appli- multiplex. All services available are an- ances or WAPPs. For example, our nounced through a specific sub-channel WAPPs provide the functionality of DAB in the multiplex so that receivers know audio encoders and multiplexers. CRC- DABMUX is used here to produce multi- where to find a service and how to decode 18 it. CrC Mobile broadcasting F/LOSS projects plexed bitstreams according to paramet- of support of mobile broadcasting on ers and content provided by the users. emerging open handset platforms and to Since their launch, the WAPPs have been catalyze an application-driven ecosys- used steadily by various members of the tem. As with Android, third party de- industry and this process turned out to be velopers would be able to create new an excellent mechanism to test our own apps long before compatible receivers ac- software. Sometimes, users inform us tually reach the market. about issues that they have. Often, we are able to fix the problems and upload up- Openmokast is a complete software stack dated software components to our WAPPs that provides a high-level application server in just a few hours. programming interface (API) for the con- trol of a DAB receiver connected to a gen- Openmokast eric computer or smartphone. It demultiplexes the received bitstream and The context for the democratization of forwards selected raw sub-channel con- mobile broadcasting at the receive end is tent to upper-layer decoding and render- different. In a time when independent de- ing applications. It currently offers velopers and users are empowered to cre- decoding libraries for standard applica- ate mobile apps on iPhone and Android tions like DAB and DAB+ audio, DMB (http://android.com) platforms, broad- (http://en.wikipedia.org/wiki/Digital_Mu cast apps development remains in the ltimedia_Broadcasting) video, Slideshow hands of a few players. More importantly, and Visual Radio. It is compatible with there are still no clear signals that digital the Openmoko FreeRunner (http://wiki. broadcast chipsets will be integrated into openmoko.org) and the GNU/Linux oper- mass market mobile devices. ating system.

The stagnation of mobile broadcasting A physical extension was built to seam- technological advances could be ex- lessly integrate a USB-based receiver and plained by the innovative and competit- its antenna at the back of the FreeRun- ive wireless communications ecosystem ner. The resulting prototype was the first that is thriving today. Several kinds of open programmable handset to integrate new wireless communications technolo- the reception of live digital radio, video gies are emerging in the quest to reach and data services with typical smart- mobile users, wherever they are, with phone functions such as mobile tele- maximum throughput. It appears as if phony, wireless Internet and GPS mobile broadcasting is standing at a junc- positioning. An Android application was ture between broadcasters and MNOs developed to showcase the usability of who are driven by distinct objectives and Openmokast in a mobile Wi-Fi broadcast business models. Broadcasters are natur- hotspot configuration. This application is ally inclined to pursue and extend their offered on the Android Market, as a GPL current free-to-air services which are open source project, and is included on monetized by public funding, licensing the CRC mmbTools live CD. Software- fees and advertising. MNOs, on the other defined-radio demodulation of the DAB hand, plan to deploy mobile broadcasting signal would represent another interest- services to generate new revenue streams ing approach to get mobile broadcasting through cable-like subscriptions and pay- reception on a mobile device without the per-view models. need for specialized chipsets. However, today’s smartphones still lack the low Openmokast (http://openmokast.org) cost wideband front-ends plus the pro- cessing power to perform heavy lifting was developed at CRC to address the lack 19 CrC Mobile broadcasting F/LOSS projects tasks like signal demodulation. Android/G1 and Openmoko/FreeRun- ner. On Ubuntu, our CELT library will Unencumbered Audio Codecs run as a plug-in for GStreamer (http:// www.gstreamer.net), the F/LOSS media Many of the tools provided on our CD are framework. Of course, none of the com- offered as F/LOSS to promote innovation mercial DAB receivers can decode CELT and new developments in mobile broad- radio as CELT is not a DAB standard. casting and DAB digital radio technology. However, some key components are miss- Unencumbered codecs have the poten- ing as the free and unrestricted distribu- tial to become widely available on most tion of standardized DAB audio codecs is types of Internet media platforms, which not possible. For these codecs, controlled happen to be the same platforms broad- distribution and payment of royalties are casters are targeting for mobile broad- required. casting reception. If this happens, broadcast radio systems may have to sup- Since audio is important in a broadcast ra- port such codecs in order to remain com- dio system, we decided to integrate a roy- petitive with Internet radio on handhelds alty-free (RF) audio option with our tools. or other media devices. Interestingly, new developments in the area of RF or unencumbered codecs were Enabling Hybrid Radio being launched at the same time. For ex- ample, the IETF had just created a new In many regions of the world, where the working group to standardize an unen- transition to digital radio appears to be cumbered Internet wideband audio co- stalled, FM radio could prevail for many dec. There is also a growing interest in the years. In this context, how will radio integration of the unencumbered Theora evolve to provide all those multimedia (http://www.theora.org) codec within the enhancements envisioned for high capa- various Internet browsers to provide a city radio systems like Eureka DAB? free default option for the new HTML 5 video tag. The Open Video Alliance (http:/ One option would consist of delivering /openvideoalliance.org) promotes this radio over Internet-streaming networks solution. instead of specialized physical layer broadcast infrastructures. While this ap- We selected the CELT (http://en.wikipedia pears to be a reasonable scenario for .org/wiki/CELT) audio codec for our im- broadband-connected locations, live and plementation. CELT is a new full-band au- uninterrupted audio delivery over mobile dio codec being developed and optimized Internet today simply does not compare for low-latency Internet applications. Its to FM. Consequently, a hybrid broad- technical features make it attractive for cast/broadband approach for radio could broadcast applications. Like other free represent a more realistic evolutionary and unencumbered technologies pro- path toward its all-digital future. Current moted by the Xiph.Org Foundation, CELT FM infrastructures would be leveraged to requires no royalties and no special licens- provide reliable and dependable audio- ing. We designed and implemented a new service components while optional multi- transport protocol for CELT over DAB and media elements,would be provided out- adapted the C library provided by the of-band through the mobile Internet and Xiph project to our live CD and Open- he Radio Data System (RDS). RDS is a mokast client applications. This library communications protocol standard for now compiles on the various platforms embedding small amounts of digital in- formation, including program informa- that we use for our projects: Ubuntu, 20 CrC Mobile broadcasting F/LOSS projects tion, time and station identification, in and application developers, have down- conventional FM radio broadcasts. loaded the CRC mmbTools live CD. Our online Wapps are visited steadily. An in- In the logical continuity of our mmbTools dependent online community of en- and Openmokast work, we have integ- thusiastic users (http://opendigitalbroad rated and developed new approaches and casting.org) is actively building a know- tools to foster the development of innov- ledge base around our tools. Our com- ative hybrid FM/RDS/Internet radio ap- pact end-to-end system triggers plications and services. One important interesting discussions about the future finding was that most of the required of mobile broadcasting whenever we in- building blocks were already available as troduce it at international events and F/LOSS projects. A functional implement- trade shows (http://www.youtube.com/ ation was integrated in a few days using crcmmb). the same USRP and GNU Radio platforms that were used for our DAB projects. While the business models for broadcast- ing are blurred by new content distribu- While many new smartphones are already tion options and by new media equipped with FM and RDS receivers, applications, it is difficult to estimate the most of these platforms do not provide of- real impact of the CRC mobile broadcast- ficial APIs to third party and independent ing F/LOSS projects. We think that by developers to control the broadcast re- democratizing mobile broadcasting tech- ceivers. For example, some new Android- nologies, we increase their chance to re- based HTC (http://htc.com) products main competitive and succeed in the have FM and RDS receivers inside them, future. although no Android API provides access to this functionality and an official An- François Lefebvre joined the Communica- droid representative has mentioned that tions Research Centre, Canada, in 1999 to there are no plans to include an API. It lead its Mobile Multimedia Broadcasting seems that handset manufacturers will be team. Since then, he has contributed to the only ones able to provide FM tuning numerous national and international and RDS decoding apps. standardization efforts and R&D projects. His recent work has focused on creating We found some unofficial mechanisms and developing open software building to access FM and RDS functionality on blocks for next-generation mobile broad- some Windows Mobile handsets. Special casting networks, devices and applica- libraries were developed by hacker com- tions. With his team, he launched the CRC munities who reverse-engineered the FM mmbTools and Openmokast open source APIs. Thanks to these developments, we software projects. He writes about the produced a portable RDS decoding soft- future of broadcasting on his blog ware library that we will use to demon- Broadcasting 2.0 (http://www.broadcast strate hybrid radio application prototypes ing20.org). Mr. Lefebvre graduated from on commercially available smartphones. Laval University in Electrical Engineering where he also completed his M.A.Sc. in Conclusion 1989. He pursued his carreer in Europe, mainly in Germany, where he worked for The CRC mobile broadcasting F/LOSS ten years as engineer in R&D laboratories projects have generated a lot of interest and as freelance supervisor of software de- among the DAB community. Several velopments on emerging multimedia and dozen DAB industry players, including Internet platforms. broadcasters, universities, manufacturers 21 Experiences From the OSSIE project

"The engineer's first problem in any design The SCA’s military and commercial relev- situation is to discover what the problem ance has given rise to a community of really is." SDR developers who can benefit from George C. Beakley SCA-based open source software.

This article briefly describes OSSIE History and Role of Open Source in SCA- (http://ossie.wireless.vt.edu), a university- based Software Defined Radio based open source Software Defined Ra- dio (SDR) project based on the U.S. De- Prior to OSSIE’s initial development by partment of Defense’s Software IC postdoctoral fellow Max Robert and a Communications Architecture (SCA, http: team of Jeff Reed’s students at Virginia //en.wikipedia.org/wiki/Software_Comm Tech in 2003, the Communications Re- unications_Architecture). The OSSIE soft- search Centre Canada (CRC) developed ware has proven useful for rapid prototyp- SCARI open (http://crc.gc.ca/en/html/cr ing by industry as well as for published c/home/research/satcom/rars/sdr/prod research and education of hundreds of ucts/scari_open/scari_open), a Java- graduate and undergraduate students based open source reference implement- and short course participants. In addition ation of the SCA . to examples of OSSIE’s successes, the pro- ject’s challenges and approaches to mitig- The OSSIE core framework was de- ating and overcoming them are described. veloped in the C++ programming lan- guage to facilitate portability to Introduction embedded platforms. OSSIE also in- cludes readily mastered tools, processing SDR is a flexible approach to radio design blocks, device interface code, and docu- that allows a radio to support new com- mentation that enable basic tasks of SDR munications standards by changing the development. OSSIE fills a niche as a free radio’s software. SDR is becoming increas- resource for SCA-based SDR research, ingly prevalent in commercial as well as education, development, and rapid pro- military arenas, with examples that in- totyping. OSSIE implements a subset of clude Vanu's AnyWave cellular basesta- the SCA sufficient to build working wave- tion (http://www.vanu.com/solutions/an forms, which can be ported to run with ywave.html), Apple's iPhone, and the U.S. commercial SCA frameworks. OSSIE is li- Department of Defense’s Joint Tactical censed under the GNU General Public Li- Radio System (JTRS, http://jpeojtrs.mil). cense (GPL) and Lesser General Public License (LGPL) and can be downloaded The OSSIE project, based at Virginia Tech, at no charge. In contrast, full-featured provides open source SDR software based commercial SCA frameworks and devel- on the SCA. The software, collectively opment tools, while providing close fidel- known as OSSIE, includes a core frame- ity to the SCA and optimization for work or infrastructure software, as well as operational use under demanding condi- rapid prototyping tools and building tions, are expensive and can run into the blocks for developing SDR applications or tens of thousands of dollars per seat or waveforms. The SCA is an open SDR archi- per copy. tecture associated with JTRS and other U.S. government programs, and has also The high costs associated with commer- been used to implement commercial cial SCA software were the initial motiva- communications standards. tion for developing OSSIE as a tool for

22 Experiences From the OSSIE project research and education in resource-lim- • more than 10 short courses and tutori- ited university settings. At the same time, als, serving over 200 participants OSSIE’s low cost has proven advantage- ous in industry. OSSIE is used for SDR rap- • three peer-reviewed journal articles, id development and proof of concept four online articles, and more than implementation, as well as to introduce twenty conference papers resulting new users, developers, or customers to from the project or using the software SCA concepts. OSSIE’s licensing is attract- ive because it allows users to customize • over US$3,000,000 in related sponsored the software for their own applications. and gift-supported research

Successes of the Project OSSIE is used by projects in industry, gov- ernment, and university settings. Aalborg The open source approach has benefited University (http://cdsr.net.dynamicweb.d SDR education, research, and develop- k/Projekter/SDR_implementering_af_In ment at Virginia Tech and elsewhere. Ac- marsat_BGAN_transceiver_baseret_p%C complishments of the OSSIE project 3%A5_SCA_standard.aspx) and Gate include: House (http://www.teknologisk.dk/_root /media/34853_5_gh_bsdr_wireless_vitae • widely distributed open source SDR soft- _09.pdf) each reported its use. OSSIE is ware being used in a proof of concept of the Government Reference Architecture • over 20,000 estimated total downloads (GRA), a U.S. government architecture of source code and ready-to-run live for above 2 GHz communications termin- DVD and VMware images als. Other studies have focused on per- formance of the OSSIE software • confirmed use by students and engin- (http://eprints.nuim.ie/1415/1/PALOMO eers at over 20 universities, companies, A.pdf), used the software to explore ef- nonprofit research centers, and govern- fects of waveform granularity (http://rta.n ment laboratories ato.int/pubs/rdp.asp?RDP=RTO-MP-IST- 083), and demonstrated and documented • eight free laboratory/tutorial exercises, porting and interoperability of wave- suitable for classroom use or self-paced forms between OSSIE and CRC’s com- study, developed with the Naval Post- mercial SCA framework and toolset graduate School (http://data.memberclicks.com/site/sdf/ sdr09-02-Singh.pdf). Also noteworthy is • additional documentation via an 80+ OSSIE’s inclusion in the OpenCPI initiat- page user and installation guide ive (http://opencpi.org).

• over 20 graduate and undergraduate stu- Challenges dents supported through work on OSSIE related projects Potential challenges to a university- based open source project such as OSSIE • more than 10 graduate theses and pro- include: jects at Virginia Tech, the Naval Post- graduate School, and elsewhere that • managing competing university, dev- have used the software eloper, and stakeholder priorities

• maintaining open communications with and among stakeholders 23 Experiences From the OSSIE project

• maintaining continuity of student fund- Students are also motivated by the oppor- ing throughout each student’s degree tunity to work on software that is unusual program and widely used within the SDR com- munity. This enthusiasm on the part of • maintaining project continuity in an en- the students and Dr. Robert sustained vironment where high turnover among the project through its early period in project personnel is desirable as part of which there was no external support. the university’s primary, educational mission The project has stakeholders that include university, industry, government users, • development, testing, maintenance, funding companies, and government and configuration management of the agencies. Interest in and preferences for software itself particular capabilities or characteristics of the software varies depending on the Addressing the Challenges – Lessons stakeholder’s immediate and longer term Learned goals, so it may not be possible to sup- port or accommodate all stakeholders As mentioned earlier, Virginia Tech has equally. Tension exists between the de- benefited from the project in terms of stu- sire for rapidly enhancing the software dents supported, degrees completed, and and maintaining stability and backward support for related research including use compatibility, supporting new operating of the software in other radio design ef- system releases, supporting a wider vari- forts. Like many universities, Virginia ety of hardware platforms, and support- Tech has a center that manages licensing ing SDR research and general wireless and commercialization of technologies communication research, all within pro- developed by its students, faculty, and ject resource constraints. As the project staff, a mission that could be seen as in progresses, it is increasingly desirable to tension with the open source approach. maintain open communication with and Over the course of the project, we have among stakeholders. This is partially kept Virginia Tech Intellectual Properties achieved through use of a mailing list (VTIP) informed of the software’s status. and a wiki that includes planned project VTIP has been very supportive of the OS- milestones, bug fixes, and enhance- SIE project and recognizes the benefits of ments. At the suggestion of some of our the open source model for the university stakeholders, we plan to initiate a users’ and the broader community, while also group to encourage and formalize this in- providing the option to negotiate alternat- teraction, possibly including periodic ive licensing for those desiring it. teleconferences.

Developers on the OSSIE project are stu- Continuity in every sense is important to dents, whose main academic priority is the project. Although challenging, effect- completion of their degrees, including ive solutions have been found. New and coursework and research that are not dir- related research projects are required to ectly related to development and en- support students through two or more hancement of the OSSIE software. years of graduate school. To date, this OSSIE’s growing capability and improved has been possible due to multi-year pro- ease of use over the past few years have jects and new projects that build on past made it more useful for thesis research, successes or explore new application increasing students’ motivation to further areas. improve the software. 24 Experiences From the OSSIE project

Maintaining project funding conducive to reporting occurs via a mailing list and the both software development and present- wiki. able, publishable research is a long-term challenge. Continuity of project staff is OSSIE benefits from use of other open maintained by research faculty, by stu- source software. OSSIE runs in Linux and dents who stay or return for additional de- has been ported to embedded platforms grees or postdoctoral fellowships, such as an OMAP starter kit (OSK). We involvement of undergraduate research- have avoided reimplementing major ers over multiple years, and by frequent functionality where possible; however, it interaction of new students with more ex- is also desirable to minimize the number perienced students. Students and of dependencies to simplify mainten- postdoctoral fellows who come from oth- ance. The current version of OSSIE uses er institutions where they have used the omniORB (http://omniorb.sourceforge. software provide another potential source net), an open source CORBA implementa- of continuity. The software, as well as doc- tion. OSSIE was recently updated to umentation of installation procedures, use the GNU Radio (http://www.gnu must also keep pace with updates to soft- radio.org) interface to the Ettus Research ware dependencies and the operating sys- (http://ettus.com) Universal Software Ra- tem to remain useful over time. This dio Peripheral (USRP), a popular low cost requires extra effort on the part of student radio frequency front end. The OSSIE developers. User reports on the project rapid prototyping tools leverage the mailing list or wiki help identify incompat- Eclipse.org open source integrated devel- ibilities and often users identify solutions opment environment and use Jython.org as well. to interface with legacy Python.org code. It is possible that future versions of the Additional challenges relate to software OSSIE Eclipse Feature will be implemen- development itself. OSSIE development ted entirely in Java to simplify mainten- began and continues in an electrical and ance. computer engineering department. As the project has progressed, we have involved Conclusion faculty and student developers from our computer science department as well as The OSSIE project provides an open computer engineers and communications source resource for SDR education, re- engineers. Some work, such as develop- search, and rapid prototyping. Develop- ment of rapid development tools and user ment and use of OSSIE has led to interfaces, does not require specialization multiple publications and presentations in communications theory or computing and several graduate and undergraduate hardware and is often more easily and students have been supported on related better developed by computer science stu- research projects. The software has been dents. Electrical and computer engineers downloaded over 20,000 times and each can now concentrate on the core frame- of the accompanying labs, suited for uni- work, development and optimization of versity or self-paced study, are down- digital signal processing components and loaded at a rate of about 1,000 per year, waveform applications, and porting the while well over 100 graduate and under- software to work with new digital and ra- graduate students and 200 professionals dio frequency hardware. Configuration have attended semester, quarter, and management is achieved using the sub- short courses that feature hands-on ex- version (http://subversion.tigris.org) revi- perience with the software. sion control system, connected to a Trac (http://trac.edgewall.org) wiki, while bug 25 Experiences From the OSSIE project

The project faces interesting challenges Jeffrey H. Reed is the Willis G. Worcester and opportunities due to its specialized Professor in the Bradley Department of but heterogeneous user base and reliance Electrical and Computer Engineering and on student developers supported by fun- director of Wireless @ Virginia Tech. His ded research projects and grants. Use of area of expertise is in software radios, cog- open source tools for configuration man- nitive radios, wireless networks, and com- agement, bug tracking, and communica- munications signal processing. He is an tion has proven valuable to the project. IEEE Fellow and the author of Software Radio: A Modern Approach to Radio OSSIE is supported in part by the National Design (Prentice Hall, 2002) and An Intro- Science Foundation under Grant No. duction to Ultra Wideband Communica- 0520418. Any opinions, findings, and con- tion Systems (Prentice Hall, 2005). clusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Stephen H. Edwards, Associate Professor National Science Foundation. of Computer Science at Virginia Tech, has interests in component-based software, automated software testing, and educa- tional uses of computers. As the PI on an NSF phase II CCLI project, he developed Web-CAT, the most widely used open- Recommended Resource source automated grading tool for com- puter programming assignments, with Why are the latest smart phones using nearly 10,000 users at over 30 institutions SDR? worldwide. He is also a member of his de- http://groups.sdrforum.org/p/bl/et/ partment's undergraduate program com- blogid=20&blogaid=20 mittee, and chair of the subcommittee on curriculum and courses. Dr. Edwards has a background in component-based sys- tems and has collaborated on software- defined radio research since 2007. Carl B. Dietrich is a Research Assistant Pro- fessor in the Bradley Department of Elec- trical and Computer Engineering at Frank E. Kragh is an Assistant Professor of Virginia Tech, where he completed Ph.D. Electrical and Computer Engineering at and M.S. degrees after graduating from the Naval Postgraduate School in Texas A&M University. He worked with the Monterey, California. Dr. Kragh received Defense Information Systems Agency, Ar- his B.S. from Caltech in 1986, his M.S. lington, Virginia and Bell Northern Re- from the University of Central Florida in search, Richardson, Texas and conducted 1990, and his Ph.D. from the Naval Post- research on adaptive and diversity an- graduate School in 1997. His chief re- tenna systems and radio wave propaga- search and teaching interests are digital tion. His current work in software defined communications, software defined radio, radio (SDR) includes leading projects re- multiple-input multiple-out systems, and lated to the OSSIE open source effort. He military communications systems. chairs the Wireless Innovation Forum Edu- cational Work Group, is a member of IEEE, ASEE, and Eta Kappa Nu, and is a Profes- sional Engineer in Virginia. 26 The Open Source Mobile Cloud

"... the cloud will soon become a disruptive This device fragmentation is poised to be- force in the mobile world, eventually be- come more difficult to address before it coming the dominant way in which mo- improves as several new types of mobile bile applications operate." devices are entering the market. These in- Sarah Perez clude e-book readers, tablet PCs, digital http://tinyurl.com/n5rsj7 picture frames, wifi cameras, and wire- less cars and appliances. These devices Cloud computing is gaining acceptance have increasing amounts of storage and as an efficient and cost-effective architec- connectivity, and are good candidates to ture to deploy many types of systems. connect to a mobile cloud. It is easy to More recently, mobile cloud computing imagine that people will want to access a has entered the scene, as an important wide variety of data and content on differ- means to deliver mobile apps and data. ent mobile devices. This article discusses trends that are driv- ing the adoption of the mobile cloud, im- For example, people may want to access portant components of mobile cloud Gmail addresses on a wifi camera or e- infrastructure, and the role of open book reader so they can email photos or source. an article. They may want to view a work or family calendar on a home refrigerator Overview or vehicle screen, or they may want to post a photo from a camera phone to a Cloud computing provides a flexible, con- corporate intranet or social network. A venient and affordable way to remotely mobile cloud should provide the infra- distribute processing and data. In the con- structure to support these possibilities text of mobile devices, the ability to off- and more. load processing and data storage is highly desirable because mobile devices have Mobile Cloud Components limited processing, memory, network bandwidth and battery life. A mobile There are several essential elements of cloud infrastructure that remotely per- mobile cloud infrastructure that make it forms computing, manages data, and can distinct from a regular computer cloud, back up wireless devices in the event of as depicted in Figure 1. loss, power failure or network interrup- tion, makes for a compelling use case. The primary purpose of a mobile cloud is to make it easy to sync mobile phones Computing requirements of a mobile and devices with systems such as social cloud differ significantly from those of a networks, email systems, computers, and regular computer cloud. The biggest dif- virtually any data store. The goal is to per- ference is found in the diversity of suppor- form processing and to manage data in ted mobile devices. There are four billion the cloud, to offload these functions from mobile phones, representing a vast array mobile devices. What follows is a descrip- of mobile operating systems and hard- tion of the significant components and ware platforms. While mobile standards capabilities of the mobile cloud. exist on large numbers of phones, there are significant variations that make it practically impossible to provide rich, us- able mobile apps that support more than a small fraction of devices.

27 The Open Source Mobile Cloud

Figure 1: Mobile Cloud Infrastructure Elements

Sync engine: a mobile cloud should be sources. This applies to mobile data such able to sync a wide variety of data and as address books, calendars and email, as content, between any source and device. well as rich media such as photos and Some people may question whether video. Consider the proliferation of syncing is still needed in an age of contacts in systems such as email, social broadband wireless networks. The answer networks, VoIP and more. People may is "Yes". Even with fast 4G networks, there have hundreds or thousands of contacts will still be pockets of non-networked in multiple places, yet they typically only areas and times when devices are offline, want a small fraction of these on their and people will still want access to their phone. A web 2.0 portal should make it data and content. Furthermore, for a easy to set up groups of users or to good user experience, it is necessary for indicate which contacts to include from many apps to access local device data. For which sources. example, users do not want to wait while video buffers. Another example is posting photos from a phone to multiple destinations, such as Web 2.0 portal: a second major element social networks, photo sharing sites or is a web 2.0 interface for user data and personal computers. There needs to be content. This provides a means to view, an easy way to allow people to specify manage, edit and filter data and content how rich media should be managed. A that flows between devices and data 28 web 2.0 portal that provides an intuitive The Open Source Mobile Cloud desktop-like user interface in a web that the change automatically propagate browser, to access and manage mobile everywhere it should, without the user cloud data, is important. initiating an update. This is the role of push notifications, which can be per- Device management: small, portable and formed using a variety of methods, in- relatively inexpensive mobile devices are cluding TCP/IP, SMS and polling. Some dropped, broken, lost, stolen and ex- networks and devices are only capable of changed with greater frequency than oth- supporting certain forms of push notifica- er computing devices. This not only tion, so the form of push notification makes it more important to back them used needs to conform to the profile of up, in case their data becomes lost, but it the involved networks and devices. makes them more costly to support. An important aspect of a mobile cloud plat- Aggregation: many mobile cloud apps re- form is the ability to remotely manage quire aggregation, such as gathering data devices over the air, in terms of provision- from multiple email systems, social net- ing devices, performing diagnostics, up- works and other systems. The mobile dating software and settings, and cloud platform should be able to intelli- remotely locking devices and erasing data gently source data from a variety of sys- for security reasons. These functions are tems. Considerations include how often typically found today with higher end remote systems are accessed and which smartphones such as BlackBerries and data is cached on the server versus stored iPhones, but they are increasingly becom- locally or pointed to remotely. ing expected with other types of portable devices. Conflict resolution: when working with data from multiple sources, one of the Data adapters: to sync a wide range of most common yet perplexing challenges data and content, there needs to be an is reconciling differences among like easy and flexible way for mobile cloud data. A simple example stems from hav- apps to access diverse systems such as so- ing someone's name in a mobile address cial networks, email systems, databases, book, while having a different version of customer resource management (CRM), their name in an email system or social and enterprise resource planning (ERP) network. When aggregating this informa- applications and servers. Without this tion, it is easy to end up with multiple ability, it could take too long to perform entries representing the same person. even simple tasks. An important compon- ent of mobile cloud infrastructure is data A critical capability is detecting 'twins' by adapters that provide the rapid ability to comparing attributes such as email ad- sync with common systems and to supple- dresses, phone numbers and other data, ment this with the ability to interface with to determine whether these are the same custom systems. person. There need to be configurable rules for determining which data should Beyond these core mobile cloud infra- win a conflict. This may be viewed as a structure components, there are several fairly arcane aspect of mobile cloud ser- additional capabilities that are important vices, but maintaining the integrity of in a mobile cloud platform, as illustrated people's data is paramount, and a robust by the smaller clouds in Figure 1. conflict resolution system is a must.

Push notifications: when data or content Core apps: many mobile cloud apps in- is changed in one place, for example on a volve a common set of functions, such as syncing contact data, calendars, email, mobile device or online, it is important 29 The Open Source Mobile Cloud files and photos. It is important for mo- mobile cloud without mentioning the bile cloud infrastructure to provide com- role of open source. mon capabilities so these functions can be performed without reinventing the Two years ago, if you mentioned open wheel. source and mobile in the same sentence, most people thought you were referring Privacy and access: as user data is stored to obscure projects for developers. Much in the cloud, it is critical that data is has changed. Google's Android open highly secure and backed up. At the same source mobile operating system has been time, there needs to be a simple way for embraced by many top device manufac- users to specify which data to share with turers and is an attractive platform for de- other people and systems. velopers due to its openness. On the server side, Funambol's open source mo- Rapid development: there needs to be a bile cloud sync server has been adopted way to rapidly build mobile cloud apps by leading mobile companies (http://fun that work on a variety of mobile phones ambol.com/news/pressrelease_2009.10.2 and devices. Until recently, developers 8.php) and has been downloaded three either needed to build native apps for million times. These projects illustrate each mobile platform, which was ex- that open source is transforming mobile tremely expensive and labour-intensive, and is attractive to both companies and or build web apps, which worked on developers. many phones but were unattractive and clunky. There are some new initiatives In the case of Funambol (http://www.fu that purport to provide developers with nambol.com), an open source mobile the best of both worlds: the creation of cloud platform, open source permeates one version of a mobile app that can be all aspects of the solution. The vast ma- widely deployed, while exhibiting many jority of Funambol code is open source of the characteristics of native apps such and there is a large, worldwide com- as a rich user interface, local data storage, munity of more than 50,000 developers and integration with other apps on the who have contributed to the Funambol device. Examples include the newly an- project. Their contributions generally fall nounced Wholesale Applications Com- into one of the following areas: munity (WAC, http://wholesaleappcomm .com) initiative, technology from 1. Device compatibility: the largest area rhomobile (http://www.rhomobile.com), of contribution pertains to enabling Fun- and a newly announced open source mo- ambol to support more mobile devices bile web 2.0 framework from Funambol. than comparable software. This stems from the community's development and Scalability: an important aspect of mo- testing of Funambol software on a large bile cloud infrastructure is the ability to range of mobile devices and sharing this support large numbers of users and, in work. some cases, millions or tens of millions of devices. This can be accomplished by us- 2. Connectors: there are several dozen ing industry standard application servers Funambol community projects for con- and infrastructure, and approaches for necting the Funambol mobile cloud serv- load balancing and fault-tolerance. er with various email and groupware systems, social networks, CRM systems The Role of Open Source and other applications.

It would be seriously remiss to discuss the 30 The Open Source Mobile Cloud

3. Server enhancements: some Funam- bol community developers have extended the core Funambol mobile cloud sync server to provide advanced functionality such as server-to-server synchronization.

4. Many other open source contribu- tions: Funambol has received numerous additional contributions, in the form of software and documentation that has been translated into different languages, significant feedback about the software performance, and much more. Funam- bol's mobile cloud platform has reached a critical mass of functionality and usage due to open source. Hal Steger is Vice President of Marketing at Funambol, inc., the leader provider of Conclusion open source mobile cloud platforms for billions of devices. Hal has over twenty Several trends are favouring the increased years of experience in the high tech in- adoption of mobile cloud infrastructure, dustry. He holds an MBA (M.S.I.A.) from not the least being the growing diversity Carnegie-Mellon University and a B.S. of mobile phones and devices. There are from the University of Michigan with a several major components of a mobile concentration in computer science. cloud platform, including a sync engine, web 2.0 portal, device management and data adapters. Important capabilities of mobile cloud infrastructure include sync as well as push notifications, aggregation, conflict management, core app function- ality, privacy and access controls, rapid development and scalability. Open source has played a major role in the evolution of mobile cloud infrastructure by signific- antly increasing device compatibility, con- nectors with other systems and many other aspects.

31 The State of Free Software in Mobile Devices

"We expect that, by 2012, around 62 Historically, the mobile user made ar- percent of the whole smartphone market rangements with some network carrier will be open source with Symbian, through a long-term contract. That carri- Android and other Linux flavours." er "gave" the user a phone or discounted Roberta Cozza, Gartner analyst it as a loss-leader). Under that system, few people take their phone hardware I started using GNU/Linux and Free choice all that seriously. Perhaps users Software (http://gnu.org/philosophy/free pay a bit more for a slightly better phone, -sw.html) in 1992. In those days, while but generally they nearly always pick everything I needed for a working among the limited choices provided by computer was generally available in the given carrier. software freedom, there were many components and applications that simply Research in Motion (http://rim.com) was did not exist. For highly technical users the first to provide corporate-slave-ori- who did not need many peripherals, the ented email-enabled devices. Today, Free Software community had reached a most people using a "smart phone" are state of complete software freedom. Yet, using one given to them by their employ- in 1992, everyone agreed there was still er to chain them to their office email much work to be done. Even today, we 24/7. still strive for a desktop and server operating system, with all relevant Apple, excellent at manipulating users in- applications, that grants complete to paying more for a product merely be- software freedom. cause it is shiny, also convinced everyone that now a phone should be paid for sep- Mobile telephone systems are not all that arately, and contracts should go even different from 1992-era GNU/Linux longer. The "race to mediocrity" of the systems. The basics are currently phone market has ended. Phones need available as Free, Libre, and Open Source real features to stand out. Phones, in fact, Software (F/LOSS). If you need only the aren't phones anymore. They are small bare minimum of functionality, you can, mobile computers that can also make by picking the right phone hardware, run phone calls. an almost completely F/LOSS operating system and application set. Yet, we have Free Software in Nokia Devices so far to go. This article discusses the current penetration of F/LOSS in mobile If these small computers had been intro- devices and offers a path forward for free duced in 1992, I suppose I'd be left writ- software advocates. ing the "Mobile GNU Manifesto", calling for developers to start from scratch writ- A Brief History ing operating systems for these new com- puters, so that all users could have The mobile telephone market has never software freedom. functioned like the traditional computer market.

32 The State of Free Software in Mobile Devices

Fortunately, we have been given a head The common joke is that you can't even start. Unlike in 1992, not every company charge the battery on your N900 without in the market today is completely against proprietary software. releasing Free Software. Specifically, two companies have seen some value in re- While there are surely people inside leasing (some parts of) phone operating Nokia who want more software freedom systems as Free Software: Nokia and on their devices, Nokia is fundamentally Google. However, the two companies a hardware company experimenting with have done this for radically different reas- software freedom in hopes that it will bol- ons. ster hardware sales. Convincing Nokia to shorten that proprietary list will prove dif- Nokia likely benefited greatly from the ficult, and the community based effort to traditional carrier system. Most of their replace that long list with Free Software phones were provided relatively cheaply (called Mer, http://wiki.maemo.org/Mer) with contracts. Their interest in software faces many challenges. These challenges freedom was limited and perhaps even increased with the recent Maemo merger non-existent. Nokia sold new hardware with to form MeeGo (http://mee every time a phone contract was re- go.com). newed, and the carrier paid the differ- ence between the loss-leader price and Free Software in Google Devices Nokia's wholesale cost. The software on the devices was simple and mostly intern- Fortunately, hardware companies are not ally developed. What incentive did Nokia the only entity interested in phone oper- have to release software in software free- ating systems. Google, ever-focused on dom? routing human eyes to its controlled ad- vertising, realizes that even more eyes In parallel, Nokia had chased another will be on mobile computing platforms in market: the tablet PC. Not big enough to the future. With this goal in mind, Google be a real computer, but too large to be a released the Android/Linux system (http: phone, these devices have been an idea //www.android.com), now available on a looking for a user base. GNU/Linux re- variety of phones in varying degrees of mains the ideal system for these devices, software freedom. and Nokia saw that. Nokia built the Debi- an-ish Maemo (http://maemo.org) sys- Google's motives are completely tem as a tablet system, with no phone. different than Nokia's. Technically, However, the market for these devices Google has no hardware to sell. They do has been minute. have a set of proprietary applications that yield the "Google online experience", and Few understand why Nokia took so long deliver Google's advertising. From to use Maemo as a platform for a tablet- Google's point of view, an easy-to-adopt, like telephone. But, a few months ago, licensing-unencumbered platform will they finally released one. The N900 (http:/ broaden their market. /maemo.nokia.com/n900) is among only a few available phones that make any Android/Linux is a nearly fully non- strides toward a fully free software phone copylefted (http://en.wikipedia.org/wiki/ platform. Yet, the list of proprietary com- Copyleft) phone operating system ponents required for operation (http://wi platform where Linux is the only GPL ki.maemo.org/Why_the_closed_package licensed component essential to s) remains quite long. Android's operation. 33 The State of Free Software in Mobile Devices

Google wants to see Android adopted Other Android/Linux devices are now broadly in both Free Software and mixed available, such as those from Motorola. Free/proprietary deployments. Google's There appears to have been no detailed goals do not match that of the software analysis done yet on the relative propriet- freedom community, so in some cases, a ary/freeness ratio of these Android de- given Android/Linux device will give the ployments. One can surmise that since user more software freedom than the these devices are from traditionally pro- N900, but in many cases it will give much prietary hardware makers, it is unlikely less. that these platforms are freer than those available from Google, whose maximal in- The HTC Dream (http://www.htc.com/ terest in a freely available operating sys- www/product/dream/overview.html) is tem is clear and in contrast to the the only Android/Linux device I know of traditional desires of hardware makers. where a careful examination of the neces- sary proprietary components have been Mobile and the Free Software analyzed. Obviously, the Google experi- Community ence applications are proprietary. There also are about 20 hardware interface lib- Whether the software is from a hardware raries that do not have source code avail- maker trying something new to sell their able in a public repository (http://trac.osu hardware, or an advertising salesman osl.org/trac/replicant/wiki/HTCDreamPr who wants some influence over an oper- oprietaryDrivers). However, when lined ating system choice to improve ad deliv- up against the N900 with Maemo, An- ery, the software freedom community droid on the HTC Dream can be used as cannot assume that the stewards of these an operational mobile telephone and 3G codebases have the interests of the user Internet device using only three propriet- community at heart. Indeed, the interests ary components: a proprietary GSM (http: between these disparate groups will only //en.wikipedia.org/wiki/Gsm) firmware, occasionally be aligned. Community-ori- proprietary Wifi firmware, and two audio ented forks, as has begun in the Maemo interface libraries. Further proprietary community with Mer, must also begin in components are needed if you want a the Android/Linux space too. We are working accelerometer, camera, and slowly trying with the Replicant project video codecs as their hardware interface (http://trac.osuosl.org/trac/replicant/wi libraries are all proprietary. ki), founded by myself and my colleague Aaron Williamson. Based on this analysis, it appears that the HTC Dream currently gives the most soft- A healthy community-oriented phone op- ware freedom based on the An- erating system project will ultimately be droid/Linux platform. It is unlikely that an essential component to software free- Google wants anything besides their ap- dom on these devices. Consider the fate plications to be proprietary. While Google of the Mer project now that Nokia has an- has been unresponsive when asked why nounced the merger of Maemo with Mob- these hardware interface libraries are pro- lin. Mer does seek to cherry-pick from prietary, it is likely that HTC, the hard- various small device systems, but its fo- ware maker with whom Google cus was to create a freer Maemo that contracted, insisted that these compon- worked on more devices. Mer now must ents remain proprietary. While no de- choose between following the Maemo in tailed analysis of the Nexus One the merge with Moblin, or becoming a (http://google.com/phone) is available true fork. yet, it's likely similar to the HTC Dream. 34 The State of Free Software in Mobile Devices

Ideally, software freedom is a community- have ignored Nokia's other release, the led effort, but there may not be enough Symbian codebase (http://www.symbia community interest, time and commit- n.org). Every time I write or speak about ment to shepherd a fork while Intel and the issues of software freedom in mobile Nokia push forward on a corporate-con- devices, I'm chastised for leaving it out of trolled codebase. Further, Moblin will the story. My answer is always simple: likely push the MeeGo project toward when a F/LOSS version of Symbian can more of a tablet-PC operating system be compiled from source code, using a than a smart phone. F/LOSS compiler or software develop- ment kit (SDK, http://en.wikipedia.org/ A community-oriented Android/Linux wiki/Sdk), and that binary can be in- fork has more hope. Google has little to stalled onto an actual working mobile lose by encouraging and even assisting phone device, then I will believe that the with such forks as its goals include wider Symbian source release has value beyond adoption of platforms that allow deploy- historical interest. We have to get honest ment of Google's proprietary applica- as a community about the future of Sym- tions. Operating system bian: it's a ten-year-old proprietary code- software-freedom-motivated efforts will base designed for devices of that era that be met with more support from Google doesn't bootstrap with any compilers our than from Nokia and/or Intel. community uses regularly. Unless there's a radical change to these facts, the code Any operating system, even a mobile belongs in a museum, not running on my device one, needs many applications to phone. be useful. Google experience applications for Android/Linux are merely the tip of I must also mention the FreeRunner the iceburg in the plethora of proprietary device and OpenMoko (http://wiki.open applications that will be available for Mee- moko.org/wiki/Main_Page). This was a Go and Android/Linux platforms. For noble experiment: a freely specified hard- F/LOSS developers who don't have a tal- ware platform running 100% F/LOSS. I ent for low-level device libraries and oper- used an OpenMoko FreeRunner myself, ating system software, these applications hoping that it would be the mobile phone represent a straightforward contribution our community could rally around. I do towards mobile software freedom. On this think the device and its software stack point, we can take a page from Free Soft- has a future as an experimental, hobbyist ware history. From the early 1990s on- device. But, just as GNU/Linux needed to ward, fully free GNU/Linux systems focus on x86 hardware to succeed, so succeeded as viable desktop and server must software freedom efforts in mobile systems because disparate groups of de- systems focus on mass-market, widely velopers focused simultaneously on both used, and widely available hardware. operating systems and application soft- ware. We need that simultaneous di- Jailbreaking and the Self-Installed versity of improvement to actually System compete with the fully proprietary altern- atives, and to ensure that the "mostly When we decided to move our office as F/LOSS" systems of today are not the close to a software freedom phone plat- "barely F/LOSS" systems of tomorrow. form as we could, we picked An- droid/Linux and the HTC Dream. We Other Systems To Consider carefully considered the idea of permis- sion to run one's own software on the Careful readers have likely noticed that I 35 The State of Free Software in Mobile Devices

our for some time. Our community device. In the desktop and server system should utilize the F/LOSS releases from market, this is not a concern, but on mo- companies, but should not forget that, bile systems, it is a central question. until viable community forks exist, soft- ware freedom on these devices exists at The holdover of those carrier-controlled the whim of these companies. A tradition- agreements for phone acquisition is the al "get some volunteers together and demand that devices be locked down. write some code" approach can achieve Devices are locked down first to a single great advancement toward community- carrier's network, so that devices cannot oriented F/LOSS systems on mobile (legally) be resold as phones ready for any devices. Developers could initially focus network. Second, carriers believe that on applications for the existing "mostly they must fear the FCC (http://en.wikiped F/LOSS" platforms of MeeGo and An- ia.org/wiki/Federal_Communications_Co droid/Linux. The challenging and more mmission) if device operating systems urgent work is to replace lower-level pro- can be reinstalled. prietary components on these systems with F/LOSS alternatives. On the first point, the HTC Dream, while somewhat more expensive than T-Mobile This article is Copyright (C) 2010, Bradley branded G1s, permit the user to install M. Kuhn, and licensed under the CC-By- any operating system on the phone, and SA-3.0 USA license (http://creativecommon extract no promises from the purchaser. s.org/licenses/by-sa/3.0/us). Google has no interest in locking you to a single carrier, but only to a single Google experience application vendor. Offering a Bradley M. Kuhn is the Policy Analyst and user "carrier freedom of choice", while ty- Technology Director at the Software Free- ing those users tighter to Google applica- dom Law Center. He worked during the tions, is probably a central part of their 1990s as a system administrator and soft- marketing plans. ware development consultant for Westing- house, Lucent Technologies, and The second point, fear of an FCC crack numerous small companies. In January down when mobile users have software 2000, he was hired by the Free Software freedom, is beyond the scope of this art- Foundation (FSF). From 2001 until 2005, icle. However, what Atheros has done he served as FSF's Executive Director, with their Wifi devices shows that soft- where he led FSF's GPL enforcement ef- ware freedom and FCC compliance can forts, launched the Associate Member pro- co-exist. Furthermore, the central piece of gram, and authored the Affero GPL. In FCC's concern, the GSM chipset and firm- 2005, he left FSF to join the founding team ware, runs on a separate processor in of SFLC. Kuhn holds a summa cum laude modern mobile devices. This is a software B.S. in Computer Science from Loyola Col- freedom battle for another day, but it lege in Maryland, and an M.S. in Com- shows that the FCC can be pacified by puter Science from the University of keeping the GSM device a black box to the Cincinnati. His Master's thesis discussed Free Software running on the primary pro- methods for dynamic interoperability of cessor of the device. Free Software languages. He is also a dir- ector and president of the Software Free- Conclusion dom Conservancy, and a member of the autonomo.us committee, which studies is- Seeking software freedom on mobile sues of software freedom as they relate to devices will remain a complicated endeav- software as a service. 36 recent reports

Online Guide to Open Access Journals Publishing

Copyright: Co-Action Publishing and Lund University Libraries Head Office

From the Introduction:

This guide focuses on Open Access scholarly journals publishing. By “Open Access journals” we refer to the publication of peer reviewed scientific manuscripts under the umbrella of a spe- cific journal title. The Online Guide to Open Access Journals Publishing is a web-based, living document that allows users to navigate quickly to specific areas of interest. Each chapter con- tains links to additional resources on the same topic in the form of: other documents and web- sites, tools and templates that can be adapted for your own use, and examples and best practices from other editorial teams to illustrate how the information can be implemented. http://www.doaj.org/bpguide/

Open Collaboration within Corporations Using Software Forges

Copyright: Dirk Riehle, John Ellenberger, Tamir Menahem, Boris Mikhailovski, Yuri Natchetoi, Barak Naveh, Thomas Odenwald

From the Abstract:

Over the past 10 years, open source software has become an important cornerstone of the software industry. Commercial users have adopted it in standalone applications, and software vendors are embedding it in products. Surprisingly then, from a commercial perspective, open source software is developed differently from how corporations typically develop software. Research into how open source works has been growing steadily. One driver of such research is the desire to understand how commercial software development could benefit from open source best practices. Do some of these practices also work within corporations? If so, what are they, and how can we transfer them? http://dirkriehle.com/2009/02/11/open-collaboration-within-corporations-using-software- forges

37 recent reports

Governance Models

Copyright: OSS Watch

From the Introduction:

A governance model describes the roles that project participants can take on and the process for decision making within the project. In addition, it describes the ground rules for participation in the project and the processes for communicating and sharing within the project team and community. It is the governance model that prevents an open source project from descending into chaos. This document explains why a governance model is necessary, considers some of the challenges associated with adopting a governance model in open source projects, and looks at the key areas such a model needs to cover. It also describes how to encapsulate your governance model in a governance document. http://www.oss-watch.ac.uk/resources/governanceModels.xml

The Open Source Way: Creating and Nurturing Communities of Contributors

Copyright: Red Hat

From the Abstract:

This guide is for helping people to understand how to and how not to engage with community over projects such as software, content, marketing, art, infrastructure, standards, and so forth. It contains knowledge distilled from years of Red Hat experience. http://www.theopensourceway.org/book/

38 Upcoming Events

April 9 April 26-28

Atlantic Canadian Entrepreneurship Expo Canada International Conference on Education Fredericton, NB Toronto, ON An unbeatable lineup of keynote speak- ers, a tradeshow, and a networking lunch The aim of CICE is to provide an oppor- make up the full-day event. Hundreds of tunity for academicians and profession- entrepreneurs come together to learn, als from various educational fields with grow, and share their success. Business cross-disciplinary interests to bridge the relationships are made, new skills are knowledge gap, promote research esteem learned, and deals are signed. and the evolution of pedagogy. http://www.atlanticexpo.ca/cities/ http://www.ciceducation.org/ fredericton

April 22

Atlantic Canadian Entrepreneurship Expo

Halifax, NS

An unbeatable lineup of keynote speak- ers, a tradeshow, and a networking lunch make up the full-day event. Hundreds of entrepreneurs come together to learn, grow, and share their success. Business relationships are made, new skills are learned, and deals are signed. http://www.atlanticexpo.ca/cities/ halifax

39 Issue Sponsor

40 Issue Sponsor

41 Contribute

The goal of the Open Source Business Re- If your answer is "yes" to any of these source is to provide quality and insightful questions, your topic is probably of in- content regarding the issues relevant to terest to OSBR readers. the development and commercialization of open source assets. We believe the When writing your article, keep the fol- best way to achieve this goal is through lowing points in mind: the contributions and feedback from ex- perts within the business and open 1. Thoroughly examine the topic; don't source communities. leave the reader wishing for more.

OSBR readers are looking for practical 2. Know your central theme and stick to ideas they can apply within their own or- it. ganizations. They also appreciate a thor- ough exploration of the issues and 3. Demonstrate your depth of under- emerging trends surrounding the busi- standing for the topic, and that you ness of open source. If you are consider- have considered its benefits, possible ing contributing an article, start by asking outcomes, and applicability. yourself: 4. Write in third-person formal style. 1. Does my research or experience provide any new insights or perspect- These guidelines should assist in the pro- ives? cess of translating your expertise into a focused article which adds to the know- 2. Do I often find myself having to ledgable resources available through the explain this topic when I meet people OSBR. as they are unaware of its relevance?

3. Do I believe that I could have saved myself time, money, and frustration if someone had explained to me the issues surrounding this topic? Upcoming Editorial Themes 4. Am I constantly correcting misconcep- tions regarding this topic? April 2010: Cloud Services

5. Am I considered to be an expert in this May 2010: Communications field? For example, do I present my Enablement research or experience at conferences? June 2010: Growing Business

July 2010: Go To Market

42 Contribute

Formatting Guidelines:

All contributions are to be submitted in .txt or .rtf format.

Indicate if your submission has been pre- Copyright: viously published elsewhere. You retain copyright to your work and Do not send articles shorter than 1500 grant the Talent First Network permis- words or longer than 3000 words. sion to publish your submission under a Creative Commons license. The Talent Begin with a thought-provoking quota- First Network owns the copyright to the tion that matches the spirit of the article. collection of works comprising each edi- Research the source of your quotation in tion of the OSBR. All content on the order to provide proper attribution. OSBR and Talent First Network websites is under the Creative Commons Include a 2-3 paragraph abstract that attribution (http://creativecommons.org/ provides the key messages you will be licenses/by/3.0/) license which allows for presenting in the article. commercial and non-commercial redistri- bution as well as modifications of the Any quotations or references within the work as long as the copyright holder is at- article text need attribution. The URL to tributed. an online reference is preferred; where no online reference exists, include the name of the person and the full title of the article or book containing the refer- enced text. If the reference is from a per- sonal communication, ensure that you have permission to use the quote and in- clude a comment to that effect. The OSBR is searching for the right Provide a 2-3 paragraph conclusion that sponsors. We offer a targeted readership summarizes the article's main points and and hard-to-get content that is relevant leaves the reader with the most import- to companies, open source foundations ant messages. and educational institutions. You can become a gold sponsor (one year If this is your first article, include a 75- support) or a theme sponsor (one issue 150 word biography. support). You can also place 1/4, 1/2 or full page ads. If there are any additional texts that would be of interest to readers, include For pricing details, contact the Editor their full title and location URL. [email protected]). Include 5 keywords for the article's metadata to assist search engines in find- ing your article.

43 Gold Sponsors

The Talent First Network program is funded in part by the Government of Ontario.

The Technology Innovation Management (TIM) program is a master's program for experienced engineers. It is offered by Carleton University's Department of Systems and Computer Engineer- ing. The TIM program offers both a thesis based degree (M.A.Sc.) and a project based degree (M.Eng.). The M.Eng is offered real-time world- wide. To apply, please go to http://www.carleton.ca/tim/sub/apply.html.

44