A TASTE OF FREEDOM
A TASTE OF FREEDOM
By Alexander Harrowell
Open-source software for mobiles is coming, and it’s not just because it’s free.
In the desktop PC world, open-source software has long ceased to be a geekish, super-minority pursuit with hippy over-tones. The spread of popular applications like Mozilla Firefox has broken down the door for free software to spread, and a computer with one open-source application is likely to gain a second.
In the world’s data centres, of course, open-source was always serious, with Unix, the various open-source BSDs, and Linux being de facto standards for the big machines, not to mention the ubiquitous Apache web server. Similarly, the programmer’s toolkit is getting progressively less proprietary, with the growing popularity of scripting languages and open-source tools such as Subversion and OpenCVS.
But mobile devices have been uncharted territory until very recently. The most common OS, Symbian, is proprietary, as are essentially all of its competitors. Comparatively little open-source application software is available, itself a reflection of the industry’s difficulties in relating to the developer community. However, it’s worth noting that a possible “mobile Firefox” gateway drug does exist – the open-source Opera browser is available as a special mobile version and is gaining a following among technical users.
Another possible source of addicts might be Nokia’s
Richard Stallman’s pioneer group set out in the early 1980s to achieve precisely this for Unix, and the result was the modern open-source movement. This requires not just open-source applications, programming languages, or developer tools, but also an open-source operating system.
Since 2005, there has been a steadily increasing number of handsets around that fulfill this criterion, essentially all being based on various Linux distributions. Now, however, a tipping-point appears to be approaching, as a succession of major actors has made decisions that imply the rapid proliferation of open-source phones. Vodafone, for example, has decided on a handsets policy that could be summed up as: “Give me standardization, but not yet.”
The carrier will in future procure devices with any of three platforms – Symbian S60, Microsoft Windows Mobile 5 (and presumably 6), or Linux. This in itself is enough to create a sizable mobile-Linux business. Given that the other two are the only mobile
On the manufacturer side, Motorola is becoming the most aggressive handset vendor in adopting Linux. Although it announced at 3GSM that it was re-opening its relationship with Symbian, and would launch a Symbian-powered handset again, it also announced a clutch of Linux devices. “Linux is our strategic platform of choice,” says Christy Wyatt, VP for Ecosystem and Market Development at Motorola. “We expect to ship 60 per cent of our product line with Linux installed.”
Motorola’s significant Chinese interests have so far accounted for the great bulk of its Linux sales. There seems to be something of a centre of expertise emerging in
E28, for example, soft-launched its 2800 series of smartphones running Linux in China in mid-2004, before bringing them to this year’s 3GSM, where it demonstrated live hand-off between UMTS and WLAN with Bridgeport Networks’ FMC solution.
Meanwhile, Yahua, makes Trolltech’s Greenphone, which was specifically designed to permit user modifications of the OS, as well as some of its own devices. Yulong, Haier, and a good half-dozen others are all at it, as are ODMs in
This raises a question – is the sudden wave of interest in mobile Linux motivated primarily by a desire to avoid paying Symbian license fees, or by a desire for greater programmability? Everyone is aware of the predictions that the next billion devices will have to be dramatically cheaper than ever before in order to reach the underserved, so the “free beer” aspect of Linux should sound attractive.
“The licensing cost is significantly less, but the work you have to do in terms of systems integration makes up for it,” says Wyatt, who adds that Motorola doesn’t intend to use Linux in its low-cost devices. “Our TTPCom products groups and AJAR are ideal for the lower end of the market.” Instead, the bulk of the product line, including top-of-the-range smartphones, will be using it, and only products aimed at specific market will not.
For example, Motorola will continue to offer some Windows Mobile devices in order to cater for enterprises that use Windows desktops and servers. “We use that platform to respond to enterprise demands,” says Wyatt. Similarly, Symbian has enough of a power-user following that it’s worth Moto’s while to have a compliant product in its portfolio.
On the other hand, Linux is Linux; the free, fully general-purpose OS used to run anything from set-top boxes and cash-points to banks’ data centres – and possibly also IMS Call Session Control Functions. There is a natural lower limit on the capabilities, and hence cost, of devices that can do it. And, whatever the device is, if it’s Linux it can be reprogrammed. There’s a reason why peculiarly masochistic geeks go round persuading unlikely electronic devices to run classic computer games, or for that matter, Windows Mobile devices to run Linux, like the Xanadux project does with HTC Blue Angel phones.
And can they be masochistic. As well as Motorola’s MobiLinux, there’s another mobile Linux distro that comes directly from the geekosystem, OpenMoko. This was developed by a group of people trying to build a mobile phone that would be open-source in hardware as well as software, which even Richard Stallman might consider over-picky. Still, all contributions are gratefully received. That’s the point of open-source software, after all.
The device, dubbed the NEO1973, is going to be produced by First International Computer, Inc. of
It will also have Push-IMAP support for push email, although to begin with, the email client wouldn’t perform. After discussions on the OpenMoko community listserv, though, one participant decided to make a new version of the client that speaks IMAP, and success was achieved. Alone among mobile devices of recent years, it won’t support MMS, after the community decided they would rather use Jabber, the open-source instant messaging protocol, or email. It will, unlike most mobile devices, have an IPSec client for high-quality secure VPN working, another reflection of the community’s priorities.
Motorola is also investing in the institutional infrastructure of mobile Linux, by setting up the LiMo Foundation as a standardization body for the technology. The company also participates in the Eclipse Foundation, where Wyatt is its representative on the board. This organization’s aim is, according to Wyatt, “to make available a full, open software development kit (SDK) for mobile Linux.”
Going open-source has, in the past, often been a way of extracting value from software projects whose owners no longer wish to support them. The popular OpenOffice package, for example, originates with the old StarOffice suite, which Sun Microsystems acquired with Star Developments in 1999 and then decided to make available as open-source software. Similarly, the old Netscape Navigator browser was open-sourced when it became obsolete, and formed the basis of the early Mozilla solution. Since then, a succession of tiring projects has done likewise.
It can be done with successful projects, too – Sun has progressively open-sourced the entirety of its Java programming language and tools, as well as the core of its Solaris operating system. The latest version of Solaris is going to be based on the current open-source version, rather than the last proprietary version – it has taken over the development process. Well-informed readers will remember that a venture capital-funded developer team was trying to build a mobile device OS written in Java, SavaJe, until it ran out of money.
Now Java itself is open-source, and SavaJe stalled, the inference is clear that throwing open the project to the developer world might just be the start of something very interesting.
Motorola’s biggest supplier for Linux phones is Oslo-based specialist Trolltech, whose Qtopia applications platform for Linux devices drives most of its product line. Adam Lawson of Trolltech claims that five million of its devices have already shipped in
“Historically, the sweet spot for Linux has been towards the high-end,” says Lawson. “Low cost handsets don’t need the power, and don’t usually have the processor grunt or the memory to make it work. The analysts who are saying that Linux is free, and hence a cost play, are wrong.” It is true that similar arguments were made by large software vendors in the IT world against Linux, Apache, and other enterprise-scale open-source projects, which have over time proven to be deeply misguided.
Lanson argues, like Wyatt and like the OpenMoko team, that it’s all about programmability. “We give our code to the open-source community and our customers so they can get their hands into the source code,” he says. “There are increasing numbers of engineers graduating with GNU, Unix, and Linux skills. More and more chipset vendors are using Linux as the initial bring-up OS.”
Mobile Linux’s biggest market is
It is in the nature of open-source software, too, that if a given market has a concentration of users it will also grow a concentration of developers, simply because anyone with the necessary clue can get cracking. Even though TD-SCDMA continues to lag, may be well worth keeping an eye (or both) on Chinese software development.
On the subject of whether the Java OS project might be revived in open-source, Lanson argues that it is a strong possibility. “Given that they tried for many years, it would be a long-term play,” he says. “But as far as I know, they were close to a fit-for-purpose software stack when they ran out of cash.”
Another Linux feature that may help it succeed in mobile is the OS’s strong support for virtualization, a legacy of its server-world beginnings. One of the Motorola Linux devices launched at 3GSM uses a single ARM11 processor core to handle both applications and real-time baseband processing, with these tasks being segregated between virtual processors. It’s one of the first single-core mobiles, as well as the first to use the ARM11 chip.
In a related development, embedded and mobile OS developer Wind River Inc. recently acquired FSMLabs, a company specializing in Linux for embedded systems. FSM’s technical approach is to run Linux as the lowest-priority task on the real-time OS, permitting a minimum of functions to be carried out in real-time and everything else on Linux. This system, RTLinux, was used in an early prototype Linuxphone FSM created with chipmaker Infineon Technologies. Co-existence with real-time operating systems, such as are used to control mobile devices’ RF functions, is a vital step in the adoption of Linux outside the multi-processor smartphone market.
More than one processor core is usually a bad idea, and multiple cores only exist in the IT world where seriously high performance is demanded. This makes it mildly paradoxical that mobile devices frequently do have them, given the constraints of battery life. The answer is, of course, the dramatically different requirements of radio-frequency and audio processing functions to anything else a smartphone might be doing. Even if that was all Linux could offer, it would already be a persuasive argument for it.
In conclusion, it’s almost surprising that Linux devices don’t already have more visibility outside
