Google and Apple’s publicized emergence onto the scene of the mobile industry is the latest consequence of the mobile phone becoming a best-selling device and a driver of growth in the telecom industry. All the big PC and Internet players want a piece of the pie, and IT industry pundits announce that finally the mobile phone will become mini-PCs. In fact, it is more likely that the PC and Internet players will have to adapt their thinking and practices in order for them to compete in the mobile phone industry.

Operating Systems 
The PC market has grown and thrived primarily due to the driving forces of Microsoft and Intel. In the mobile space no such duopoly exists. There are a wide range of operating systems available. Broadly speaking, operating systems are divided into “open” and “proprietary” operating systems. The de-facto definition of open operating systems in the mobile phone space is quite different from the PC space where “open” often equates to “open source.” In the mobile phone market, “open” refers to operating systems that have published native APIs, provide a development kit, and allow some sort of installation of applications post-ship. Following this definition, the following open operating systems are present in the market:

• Symbian OS
• Microsoft
• Linux (although the APIs are not generally used for installation of application post-ship)
• OS-X (although Apple has not yet given access to the native APIs)

These operating systems were expected to ship around 145 million units in 2007 with Symbian OS making up well over half (Richard Winsor, Nomura, Oct. 2007). While this seems impressive compared to the number of PCs shipped every year, it is only around 13% of the mobile phone market. The remaining 950 million are based on proprietary operating systems. The key proprietary operating systems are:

• S40 (Nokia’s proprietary operating system)
• DMSS with Brew (Licensed by Qualcomm together with Qualcomm chipsets)
• OSE (Licensed by Ericsson Mobile Platforms [EMP] together with EMP chipsets)
• P2K (Motorola’s proprietary operating system)

There are no independent volume forecasts for these operating systems, but we estimate they probably make up more than 650 million of the 950 million mobile phones based on proprietary operating systems.

Due to the broad availability of different operating systems, Java has become a very popular programming language supported on most phones in the market. The mobile version of Java is called MIDP and implements a limited set of APIs. Due to this limitation in available APIs and some fragmentation caused by the standardization process, Java is mainly used for games downloaded via service providers’ portals. There is no serious market for other types of applications.

While the PC market has converged on one dominated platform, signs in the mobile market indicate that it will evolve into distinct segments where different operating systems will play their role. At the high end of the market will be the open operating systems running on smartphones. These smartphones will be sold into the Enterprise and into the multimedia segment (music, digital photography and video). In the mid-range market, there mainly will be capable proprietary operating systems providing users with a broad range of applications and services. At the low end of the market, there will be inexpensive, voice-only phones with limited capabilities.

To be successful in the mobile phone industry, PC and Internet players need to understand and embrace this complexity, which involves a large investment in resources both up front and ongoing. But the fragmentation of the mobile platform doesn’t end there.

Software Storage and Execution 
In the PC and Internet market, there is a long tradition to rely on Moore’s Law to solve performance and storage issues associated with new applications. As long as the functionality and features are superior to the competition, hardware improvements to processor speed and memory availability solve performance and cost issues over time.

In the mobile market, this is not the case. Although the memory architecture for high-end smartphones has started to look like PCs and processor speeds are improving gradually, mobile phones are constrained on two fronts. First, mobile phones are battery powered, so although processors are getting faster according to Moore’s Law, battery technology is not able to keep up. If the processors are run at full clock-speed, it is impossible to maintain a reasonable time of usage. Second, mobile phones are much more price sensitive than PCs, especially in the mid- and low-end of the market. In this part of the market, you cannot double the available memory with every phone release. In a market where the sell-in price is around $70-140, you would soon price yourself out of those segments.

The mobile phones of today barely resemble those of old. For early mobile phones, the majority of the software load was responsible for running the signaling stack. It was small and could be held in a limited amount of NOR flash ROM. The benefit of NOR flash is that the code can be executed in place – hence there is no need to copy the code to RAM first. NOR flash therefore saves cost (as RAM is relatively expensive) and provides good performance for real-time execution for voice calls. NOR flash however has the draw-back that it is comparably slow at writing to memory.

We now have a situation where the storage medium used for handsets is very different across the high to low end. At the high end, there is no doubt that NAND flash will dominate and there will be plenty of it. NAND offers a good balance of characteristics including power consumption, density, cost, and functionality. These constraints do not exist entirely in the PC world, which evolved much earlier and typically now have a simple BIOS held in flash ROM and a large binary OS image stored on spinning disks. Spinning disks have many benefits but consume more power than a solid state medium such as flash ROM and are not generally as well packaged as the solid state equivalents.

As we move to the mid-tier and low-end handsets, it is more likely that NOR flash will be used. It enables the mobile phone manufacturers to hit the right price points with a limited set of functionality. As the market develops, and the battery-constrained Moore’s Law starts to kick in, more and more NAND memory will migrate to the cheaper phones.

In summary, those wishing to enter the mobile phone market must pay careful attention to the key user requirements and architect their software and hardware solutions to give the optimum user experience for the target segment.

Distribution of Software
In the PC and Internet market, companies are used to having Windows as a common native environment and the browser as a convenient extension that can both abstract the native APIs and also act as the initial window for offering new applications and services to the user. Once the user has installed or signed-up for a particular application or service, the application or service will have upgrade capabilities built in. This allows Internet and PC players to up-sell new features and functionality and to maintain the customer relationship through the life cycle of the product or service.

In the mobile phone market, things are not as easy. There is no common operating system to program against. Many handsets contain a Java Virtual Machine. This allows new functionality to be downloaded to the device over the air, but provides only a limited execution environment. This has limited the use of Java applications to mostly games. 'Open platforms' such as those using Symbian OS or Windows Mobile do allow the user to download native applications to the device. Typically both of these approaches do not allow the core functional applications to be enhanced or replaced.

Most handset vendors have also implemented a Firmware update Over-The-Air (FOTA) solution. According to a recent study from Ovum, nearly 400 million phones have shipped with FOTA to date. By end of 2009, 85% of all phones are expected to have FOTA embedded. This allows handset vendors or service providers to make updates available to all parts of the phone software over-the-air.

Both firmware update and application installation techniques can also be performed via a cable. When a manufacturer has very large images and no mature delta update / installation software, this may be the only solution available to customers to update their phones. This is the case with Apple, which has based its firmware update solution on the iTunes / iPod model. This is due to the extreme size of the software in the iPhone devices, rumored to be in excess of 500MB.

A way of merging the benefits of application installation and firmware update can be realized through implementation of an emerging open standard known as SCoMO (Software Component Management Object), defined by the Open Mobile Alliance Device Management working group. With this enabler, both applications installed to open phones, Java applications, and embedded firmware components can be remotely managed (installed, stopped, started, etc.) in a standardized way. This should help to reduce complexity and increase the distribution of all types of software to mobile handsets.

Distribution of software in the mobile phone market is more complex than in the PC and Internet world. To be successful in the mobile phone market, the new players need to understand which segment of the market the company wishes to address. If the focus is purely on the smartphone market, it is appropriate to stay with native programming and distribute the particular application via the various operators and independent portals. If the business case warrants a broader penetration, Java has to be taken into account and companies will be required to work directly with the handsets manufacturers to get the software embedded into the handset.

As we move forward with SCoMO, there will be new cross-platform technologies available that will make distribution of software much easier. To influence the development of these technologies, participation in standards organizations, like the Open Mobile Alliance, can be advantageous. 

Summary
The mobile phone market is much more complex than the large PC and Internet players might like. It is unlikely that mobile phones will resemble PCs in the next 3-5 years. As we have shown above, high-end smartphones have some characteristics in common with PCs in terms of memory availability, openness of APIs and the distribution of software. However, even in this smartphone segment, there is no single operating system to deploy applications and services on. When entering the mobile phone market, PC and Internet players need to take this complexity into consideration. With the entrance of Google and Apple into the mobile market, we now have even more competition among different operating systems.

Internet and PC players should also be very careful to not rely on Moore’s Law to solve any performance or memory issues. In the mobile phone market, the slow evolution of battery technology keeps a lid on how fast processors can be clocked. The price pressure on the mid- and low-tier phones also makes it difficult to double the available memory every time a new phone is launched. If this part of the market is targeted, writing efficient code can make a big difference.

Finally, the distribution of software is significantly different in the mobile market. It is important for new players to consider how wide a market space they would like to cover. Based on this assessment, programming language and distribution channels need to be considered. Where the business case extends all the way to the mid and low end, PC and Internet companies will soon realize they no longer have direct access to consumers and will be required to build up relationships with the handset manufacturers to get their software embedded at launch.