Jason Whitmire

How Industry Software Expertise is Helping to Solve the Android Fragmentation Dilemma

It is no secret:  While Google continues to push for wide adoption of smartphones driven by its open source Android platform, most industry pundits know that manufacturers typically need upwards of a full year of development and productization effort (and often Google’s help) to develop an Android phone. Not surprisingly, it takes much longer for an OEM to reach a level where they can irrigate the operator ecosystem with differentiated phones based on Android (this of course applies to most new mobile device software stacks). At the same time, we are all aware that there are multiple parallel efforts to enhance Android that have led many observers to believe that fragmentation is inevitable. Ultimately, it is said, this will become the bane to developers who will struggle to reach an economy of scale on a moving target where multiple variants of the Android OS co-exist.

That said, a handful of companies are seeking to close the gap in Android fragmentation diversity by providing the ecosystem with tradecraft skills, tooling and testing that allow for an unprecedented degree of platform validation that can act as a common denominator on the latest versions of Android for the many combinations of open source configurations, IP changes, and hardware revisions. Indeed, one of the biggest advantages of the diversity in Android has been that manufacturers are not tied to the success or failure of an individual hardware platform, meaning that today there are dozens of hardware programs in downstream Android development.

This, when overlayed with Google's official Android versions and integrated with the multiple moving parts of open source code and 2nd party/3rd party subsystems and applications, has created a "thicket" of software code and hardware that is difficult enough to manage inside of one manufacturer, much less between device makers and developers. Indeed, there is a growing recognition that companies are challenged to cost-effectively create a commercial handset product based on something that's inherently unstable or changing too rapidly, especially for mass market. 

Wind River is one of the companies with the deepest credentials in the Android compatibility space, having been Google's lead Linux commercialization partner for Android starting in early 2007 and having earned a seat at the OEM and operator support table by creating the most comprehensive Android “test and fix” program in the industry.  This has included deep modifications that conform with the OHA governance model as well as global support and maintenance to ensure reuse across a true Android lifecycle management program (in fact Wind River just announced its latest commercial Android platform optimized on the TI OMAP3 Platform). 

Although there have emerged a long tail of professional services software companies that have announced OEM support programs for producing Android phones over the past two years, few appear to have recognized "the pain" that open source causes the manufacturer ecosystem and why a commercial platform eases that pain. They miss the opportunity to solve the business risk problem of open source by increasing predictability – many times the key issue with open source.  Using a commercial platform, Wind River is offering that predictability, but without losing the accelerated innovation rate that a dynamic platform like Android is providing the market.  Only time will tell which approach will prove out.

Operators return to Barcelona, (this time) armed with Open Source

Scene Setter
26 years after GSM was created to design a pan-European mobile technology, Mobile World Congress number 13 is set to take place in Barcelona in February. This time around, as they did when GSM World Congress was first held in Madrid in 1995, mobile network operators will dominate the scene. 

Next month, however, the topic of discussion will not be new network deployments, or the latest traunch of jazzy new devices, or the next best application. Rather, Open Source will be topic Number 1 on the operator agenda in 2009. As changing operator strategies include the need for a strategic terminal platform that they can influence, the tectonic plates that once defined how a device was created and deployed are shifting and fueling significant change in the value chain.

A Brave, New Ecosystem
Whether it be the proliferation of phone development activity around Google’s Android stack, the phenomenal operator gravitation toward the LiMo Foundation, or Symbian’s intriguing announcement to open source its end-of-life cycle stack, the mobile industry is breaking out of the traditional controlled development environment to favor collaboration that accelerates innovation. The use of open source software in mobile is exploding from the operating system all the way up to the user experience, and Linux-based open source stacks are moving well beyond alpha stage with backing by industry heavy weights.

Indeed, the ubiquity of open source is causing not only a fundamental shift in proprietary OEM software deployment, but also accelerating the opening of the operator walled garden. As VisionMobile’s Andreas Constantinou – arguably the industry’s most astute pundit of open source in mobile – has pointed out, strategic “shared” core software platforms are “in” (versus supporting and maintaining up to ten proprietary stacks in an operator portfolio).

Innovating on a Perceived Commodity
Given that almost every major operator has now launched some sort of Linux device, operator versions of open source stacks are entering the optimization phase, aiming to reduce costs of bringing new Linux devices to market, more rapidly evolve on-device and network-based services, and speed application testing and certification. These highly optimized stacks will allow the operators and their OEM and ISV partners to shift the focus of innovation from the baseline software to the applications, services, and user experiences that will ultimately provide the basis for differentiation and subscriber and revenue growth.

At the same time, the complexity of open source coupled with proprietary software assets has forced operators to quickly get smart – few companies in mobile offer the indemnification, quality metrics, warranties, and SLAs that were intrinsic to the fully proprietary software paradigm. Yet this is exactly what the operators and their partners seek to allow them to overcome the last business hurdle to open source adoption.

Take-aways, Before You Reach Costa Brava

• A number of trends (community content, me-portals, WiFi, etc.) will accelerate the crumbling of once venerable operator walled garden.

• Operators recognize that the mobile device software experience must improve across devices, with a critical need to bring all-internet experience to mobile devices.

• Chipset manufacturers will continue to rapidly embrace open source stacks (Android, Limo, Moblin) as service providers send downstream requirements that reward pre-optimized hardware-software combinations. Expect multiple commercial devices and demos on all three software platforms in Barcelona.


• As memory and CPU constraints disappear, open source will allow mobile device to become true application servers, not just smart mobile phones.


• Fragmentation in mobile device software is out, shared core software platforms are in.

• Nevertheless, fragmentation of developer ecosystems (LiMo, Android, Moblin, Symbian, Blackberry, iPhone, Palm…) will constrain the pace of innovation in the mobile market.


• Operators will be making fundamental decisions about which stacks to deploy (Android, LiMo, Moblin, Symbian, etc.) while larger ODMs shift their attention to learn open source.

• Operators will increasingly require commercial solutions that protect signature applications while leveraging open source’s innovation rate

Have fun !

An Industry and Technology Revolution is on the Horizon

Limo, and in a slightly different way Android, have killed the standards-based approach to open source development in mobile. In the Linux world, creating an esoteric, theoretical application standard not based on market-driven code requires too much speculative investment without any clear mitigation of ROI risk for anyone to take up anymore. Indeed, the days of a bunch of representative techies flying to exotic locales to dream up the theoretical perfect system are over. It’s just too expensive to completely retool an entire stack without a known intrinsic return.

Because of the concentration in the mobile market (83% of handsets manufactured by five companies), when market leaders have invested in a stack, it is a standard whether certified by an arcane standards body or not.  The  Open Handset Alliance has created a de facto market standard not because a group of market leaders have adopted the standard, but because of Google’s overall singular market weight. The effect is the same (Trolltech and OpenMoko did the same thing that OHA did, and took it one step further by actually building a phone, but no one came running to embrace their reference designs because they lacked the market weight that Google has).

The Check is in the Mail, Do No Evil and Other Matters of Trust

All will agree that mobile Linux suffers from fragmentation today, and that fragmentation
•    creates significant challenges to Linux adoption in mobile phones (lack of interoperability et. al.) 
•    presents barriers to innovation
•    increases the carrier cost to Linux terminal deployment.

For Linux to succeed in the mobile market, we need to minimize fragmentation and its resulting incompatibility.  In the early days of the PC revolution, this type of incompatibility was similarly rampant, and the Wintel monopoly provided the standard.  More importantly, the Wintel monopoly created the incentive to rally around the standard. 

But suggesting that the industry set up Google as the mobile Linux gatekeeper, including issuing all the keys to the various feature phone middleware/applications framework kingdoms (which account for about 90% of all phones deployed today), to take on the monumental role of software guardian for mobile Linux, is possibly antithetical to the open source movement, regardless of Google’s motto.  And perhaps this is not the role that Google ultimately seeks in the mobile market.  Indeed, there are multiple thrusts to Google’s mobile terminal strategy, all of which are underpinned by the principle of radically improving the efficiency and experience of the end-user’s mobile internet time.

Open Handset Alliance and LiMo – Why both initiatives are here to stay

It's easy to take the fragmentation of the mobile phone industry for granted. With over 40 different proprietary and Open Source Middleware and Application Frameworks addressing Ultra Low Cost, Entry, Feature and Smart Phone market segments, software costs have exploded and are expected to hit some 25% of the entire Bill of Materials on a mobile phone this year. No wonder mobile network operators are asking the industry to stop the madness and consolidate around a limited number of pre-defined platforms.

Many operators very clearly see Microsoft, Symbian and one to two versions of Linux as the only future software platforms that will comprise their terminal roadmap.  Linux is unique in that most Tier 1 OEMs – with the exception of Nokia – plan to gradually replace the proprietary software addressing the feature phone segments (where by far the most volumes of phones are shipped today and in the future).

With this in mind, two juggernauts – Open Handset Alliance and LiMo – promise to alter the landscape for the third software platform – Linux – for good. The LiMo Foundation’s goal is to defragment the Linux handset market by creating a Linux-based mobile platform that lowers development costs, increases flexibility, and yields differentiated devices. The goal of the Open Handset Alliance is to deliver solutions to enable complete open software, open devices, and an open ecosystem. Are these movements destined for perpetual conflict?