Java to dominate consumer electronics?

Original URL: https://www.theregister.com/2004/04/22/java_to_dominate/

Yes, if it's lean and mean

Posted in Networks, 22nd April 2004 11:38 GMT

Java and Linux could dominate consumer electronics, but to do so, Java needs to be less fragmented and highly resource efficient, making it suitable at last for the mass market.

As the phone, PC and media center converge around the same architectures, companies from all three sectors are adapting their technologies to span the others. Microsoft is seeking to do this with Windows Mobile and real time CE, and Nokia has its phone-based home entertainment devices. Two platforms, however, already straddle these three worlds - Java and Linux. These could form a powerful combination to dominate the consumer space, rendering the mobile operating wars irrelevant outside the enterprise sector.

After years of seemingly failing to notice just how valuable an asset it had in Java, and missing many opportunities to exploit it, Sun now has its best chance ever to control a potentially universal technology. And the pressure to open source Java has eased somewhat with the signing of its recent pact with Microsoft to end patent and antitrust actions - one motivation was surely to bolster Sun's alliances and not force it to fight on two Java fronts, with IBM over open source and with Microsoft.

While the combination of Java and mobile Linux could become even more attractive to the cost-sensitive commodity handset industry with the lowering of prices that open source Java would bring, it could also increase problems of fragmentation.

This has been a big enough issue for the mobile version of Java even in its semi-closed state, and one that Sun is now seeking to address. On its other big challenge, efficiency, however, the initiative has been firmly grabbed by a third party, Java platform supplier Esmertec, which claims the slimmest implementation of the software, suitable for low cost handsets with limited memory and processor power.

Resource efficiency

For Java to live up to the forecasts that by 2008 it will ship on 85 per cent of new handsets - it is currently supported on less than a quarter of phones, though on almost half of new shipments - it needs to be highly resource efficient. To date, it has only had a place on high end smartphones and media phones, which command sufficiently large price tags that manufacturers can be generous with processor speed and memory and so counter the effects of battery drain. This has created a particularly attractive market in Japan and Korea, where consumers will bear far higher phone prices than elsewhere.

But in the European and US mass markets, price and battery life remain the top two factors in consumer choice, which means making Java efficient on small devices - while still being able to deliver the kind of services that are currently the preserve of the premium phones, but will increasingly become expected on all models.

"The demand on memory, battery and clock speed is growing more quickly than the efficiency of the hardware, so the efficiency of software becomes critical," commented Christophe Francois, head of strategic marketing at Esmertec. Yet it is vital, particularly for the success of 3G, that complex handsets that are easily customisable and offer multimedia services can be delivered at low cost, to encourage users in mature markets like Europe to upgrade their handsets and move to 3G services.

This dilemma is the basis for the main strategic thrust at Esmertec, maker of Java runtime environments and application platforms for chipmakers, OEMs and ODMs. While its main rival, Japan's Applix, focuses on feature-rich but resource hungry implementations for the customised handsets of the Asian carriers, Esmertec is aiming for a broader base and is even adapting its technology for set-top boxes, cordless and VoIP phones, machine-to-machine communications and, potentially, for in-car systems.

It has signed a deal with Time Warner for interactive TV boxes in the US, and will soon announce a telco partnership in Europe. In M2M, it is partnering with Wavecom. All of these are markets where the devices need to be low cost and compact, and so require high levels of resource efficiency, and where the dominant OS will increasingly be Linux.

The danger of fragmentation

An efficient Java-Linux combination would be a powerful candidate to provide the standard basis for consumer devices from phones to automotive systems, but there is another weakness that currently holds J2ME back from achieving its full potential - its tendency to fragmentation.

Uniformity is important for developers seeking the broadest possible base for their applications. It will also become an issue for users if they do as Nokia hopes and start running multiple phones for different purposes - portability of applications between them will then be more critical than in the current closed world.

Currently, many large handset makers have their own Java implementations that are tightly integrated with and optimised for their designs, as Nokia and Symbian do. But most have got behind the push for greater uniformity through common APIs, seeing the advantage of attracting developers and boosting the overall market for Java phones, which currently command a higher margin than other models.

For instance, Sun joined the major handset makers earlier this year to launch the "Java Verified" program, which is designed to enable Java mobile apps to run on phones from multiple vendors with only one certification required. Sun will manage and operate the scheme, and the certification will be based on the work of Java Technology for the Wireless Industry (JTWI), a group within the Java Community Process that seeks to plan future directions for mobile Java.

The JTWI, which is dominated by Sun and Nokia, aims to lay out an overall architecture for J2ME in mobile phones, which is still incompletely standardised, identifying configurations, profiles and APIs, such as MIDP 2.0, that are essential for compliance. This will not necessarily be welcome among the operators. They tend to favour Java because it does not make them dependent on a single supplier, but they are used to operating in high-walled gardens and keeping their particular platforms tightly closed. With Java becoming a tool for differentiation of applications and services, they are seeking to turn it to their advantage as the Japanese and Korean carriers have always done.

The Japanese operators have all developed their own sets of Java extensions to support their highly differentiated handset specifications, and Vodafone has now taken a leaf out of its Japanese subsidiary's book, using that company's extensions, JSCL, as the basis for the Vodafone Service Class Library, which under-pins Vodafone Live!

Sun's initiatives

Sun has made several moves recently to rein in such behaviour. The latest version of J2ME's Mobile Information Device Platform (MIDP 2), is more all-encompassing than its predecessor, leaving less leeway for the proprietary extensions that have plagued mobile Java. MIDP 2 should be standard in new handsets from midyear.

Sun argues that, as the work of initiatives like JTWI increase the base functionality of J2ME, there will be less requirement for extensions. Device specific API extensions have proliferated on mobile phones because the MIDP profile is targeted at all mobile devices and so omits many phone-specific requirements, such as address books and particular user interface behaviour. But the claim ignores the politics - the operators' all-important strategy of gaining control of the handset platform.

It also skates over the primary carrier requirement from any architecture - that it supports flexibility of design so that customised models can be cheaply and easily created. While Java-based technologies such as Esmertec's new component-based applications environment make J2ME more suitable for this purpose, there is the risk that, with JTWI, Sun is trying to create too generic a platform for mobile phones ranging from games-optimised devices to business-focused PDA hybrids.

Do phones need an OS outside the enterprise?

Despite these hurdles, the demand for rich media applications and functions such as downloading will undoubtedly drive Java, which is starting to prove itself as a strong platform to deliver such benefits, and as it becomes dominant it will shift the politics surrounding the handset operating system.

Both Symbian OS and Windows Mobile, in their current incarnations at least, are heavily geared to the smartphone - a business-oriented device that seeks to behave like a PC in a phone casing, with dedicated applications processors and an OS and interface that works similarly to that of a computer.

There is no doubt that the fight to the death between Nokia and Microsoft to control this platform is largely an enterprise issue. It is a vital one - as the smartphone evolves to become the de facto corporate client platform, slowly eclipsing the conventional PC, Microsoft needs to retain its influence by ensuring that the phone is basically another incarnation of the Wintel PC, while Nokia sees its opportunity to supplant the Windows giant and control the client architecture itself, with all the enterprise business that could leverage.

This market is peculiar because the carriers have so little power in it. But back in consumer land, the operating system picture is far less clear cut and the Java-supporting carriers hold many key cards. The application processor, which is essential for Symbian OS or Windows to run, becomes less important and the focus is on media chips and screens.

Since these are power hungry themselves, there is a strong case for omitting the apps processor and the complex OS and running Java directly on the real time operating system, present in all phones, that controls the actual running of the device. Increasingly, that cutdown OS is likely to be Linux, although Microsoft is pitching for this market too with real time implementations of Windows CE.

To become truly universal, Symbian OS will need to shrink down significantly, and command lower royalties - something the Symbian group claims will develop over the coming year as the system starts to appear in midmarket phones.

However, there is a question mark over how far there will be a place for the OS further down the food chain, where Java, once it achieves critical mass, will provide the universality that was one goal for Symbian OS. As Sun chief Scott McNealy put it: "It's so 'last millennium' to write to the operating system. You must write to the Java web services layer."

Such an approach, combined with a cutdown OS such as real time Linux, not only promises to help deliver multimedia service and customised handsets with low power consumption, but it enables the operators to avoid technology controlled by Nokia. As such, it is sure to be taken very seriously by the carriers, an opportunity that Sun and the Java industry need to seize as aggressively as possible.

Wireless Watch is published by Rethink Research, a London-based IT publishing and consulting firm. This weekly newsletter delivers in-depth analysis and market research of mobile and wireless for business. Subscription details are here.