As applications become more complex and the number of devices continues to rise, the issue of how to develop and make the application work well is high on the agenda. Andrew Darling investigates.
curious dilemma lies at the heart of European mobile operators’ ambitions when it comes to application development and the launch of new data services. The meeting of ‘smart’ mobile devices and the higher bandwidth mobile networks is creating a class of sophisticated data services, such as shopping, dining, entertainment and enterprise applications, which mobile subscribers are keen to use but unable to discover. The potential of these services, however, remains largely unrealised, leaving mobile operators to face a rare situation where content supply and user demand are fast emerging, but the tools necessary to connect the two remain inadequate. The primary reason for this is device fragmentation.
Because of the level of fragmentation in the market, developers must write code to address device-specific APIs, memory management and performance issues, localisation issues, screen-size variations, custom extensions, implementation issues, as well as operator-specific requirements. The result is that developers can spend more time manually porting applications across a variety of devices and OS to ensure they operate, rather than developing new applications. And operators struggle to find a solution that enables them to distribute new applications to their customers in a way that meets the demands of efficient development, good user experience and rapid time to market.
“The main reason for the complexity of mobile data services is that the application must cope with the bewildering diversity of devices, operating systems, versions and protocols that exists in the mobile world,” says Paul Nerger, VP sales and business development for Argogroup. “And then, must be able to work over multiple, disparate, proprietary networks.”
Fragmentation can be regarded as the evil twin of differentiation, the term that marketers use to explain the need to create so many different handsets. Mobile phones are considered very personal devices and the consequent marketing of these devices has led handset manufacturers to produce something for almost every possible taste.
While differentiation is good for manufacturers, it has become a major problem for developers. Fragmentation drives up the cost and time it takes to get an application to mass market. They have to be ported from one device to another, translated into different languages, and increasingly customised for specific mobile operator needs. The result is usually hundreds of different stock keeping units (SKUs) of the same product.
This issue does not apply so much in the case of smartphones, such as those running on Microsoft, Symbian or Palm OS platform, but it is a major headache for mass-market feature phones. Eleven million smartphones shipped in the last quarter of 2004, according to Strategy Analytics — an increase of 20 percent on the previous quarter — but even though this represents a growing market share, the majority of subscribers use mid-range ‘smart enough’ phones and no single operating system dominates this sector. Nokia, with its 35 percent market share, has its own operating system and a vested ownership of Symbian, yet it runs multiple operating systems across its device portfolio.
Mobile operators can exert more control and influence over the applications access by these mid-range feature phones but as already mentioned, even for a J2ME developer, the fragmentation of devices creates significant obstacles.
To address these issues and the changing requirements of different markets and regions, handset manufacturers have begun looking at using multiple baseband platforms to serve different application requirements.
“The problems that handset makers have are caused by operator strategies now being deployed in Europe, says Steven Baker, sales director for TTPCom’s software business. “Strong operator positioning means unless you adhere to their portal schema and play their game, your strategy becomes ad hoc rather than meaningful.”
A platform-based approach, such as the one advocated by TTPCom and its Ajar platform, recently licensed to Motorola, allows an entire set of pre-integrated and pre-tested applications such as phone book, organiser, image viewer and screensaver, to be re-used and customised for multiple phones over different networks.
“For example, Vodafone specifies two separate inboxes for SMS and MMS delivery whereas T-Mobile only wants one. How do I do the implementation when it requires such code divergence?” says Baker.
Using the Ajar platform, handset manufacturers can select best-of-breed third party applications, adding in-house functions to increase diversification. The tools made available by TTPCom enable application development, user interface (UI) customisation and product test and validation in a way that shrinks development cycles, according to the company. Following TTPComs’s model of platform-based application development, PC-based simulation of network activities and hardware events can be re-created which should improve development productivity.
“You need one development organisation serving as many cells in the development cycle as possible and this is where we are going with the Ajar platform,” says Baker. “We are dealing with handset makers primarily but we still talk to operators who can use the tools for the language translations. The tool set which comes with the application framework can be used for the localisation and customisation by operators via a PC and then sent as an XML package to the manufacturer.”
Analysts, however, remain less convinced that developing from a baseband chip level enables the flexibility required by operators, as Ovum’s Bola Rotibi explains.
“Writing direct to the chip obviously offers potentially huge performance advantages but lacks flexibility in terms of updating applications, unless you can update device firmware over the air (OTA). While products exist, there’s not much evidence that operators are doing this at this time.”
Rotibi believes developing on the middleware platform probably offers the best in terms of flexibility, providing the middleware itself is sufficiently good and the hardware performance is up to the job. One company that is tackling the problem from this perspective is Intuwave and its CTO, John Hoskins, says the company is founded on the premise that it is difficult to develop applications to mobile phones.
“APIs are not always published — even with Symbian devices — so there are always going to be problems in migrating applications from one device to another,” says Hoskins. “But middleware platforms provide basic APIs and the tools necessary to avoid each individual OS API on devices.”
Intuwave’s core proposition is to provide a cross-platform integration engine called m-Network, a middleware layer, which allows application developers to build applications within a familiar PC environment or scripting language that can be deployed across multiple devices.
However, Intuwave’s solution only serves the smartphone market and while operators do face issues in managing smartphone devices in a way that doesn’t incur high customer support costs, a bigger slice of the market concerns mid-range feature phones. But in a move it describes as “shattering the mobile usability barrier”, Seattle-based Action Engine announced last month that it plans to bring its Mobile Applications Platform and Brand-n-Go Applications Pack to Java-enabled mobile phones.
The Java version of Action Engine’s smartphone middleware platform is being built using Java MIDP 2.0 specifications and will be available in September 2005. When completed, the Java version of the mobile application platform will be small enough in size to be provisioned OTA to a mobile device, according to Ted Wugofski, CTO of Action Engine.
“The Java platform will be great for developing,” says Wugofski. “Series 40 to 60 handsets performed without much change when using our platform. By expanding our addressable market base through support for Java devices, we can offer mobile operators the ability to give millions of frustrated data services users a better alternative for accessing content wirelessly.”
Wugofski says that most operators didn’t want to deploy Action Engine’s original platform because it only addressed the Symbian and Microsoft market sector and needed something that could target the far larger Java sector. “There was a big change at this years’ 3GSM World Congress with phone manufacturers wanting third party applications themselves to add more value for both operators and enterprises.”
Using an open mobile data services architecture, Wugofski says operators can strike at the heart of the mobile application dilemma. Such an approach relies on client software in the device, working in concert with server software, to deliver an easy-to-use and intuitive application.
Both the mobile device and application market are alive with innovation, both technical and lifestyle. As applications become more complex and the number of devices continues to rise, the issue of how to develop and make the application work well on the handset will continue to remain prevalent. A Java run-time engine in the phone will process XML feeds provided OTA from the server side, thereby integrating these two elements and enable a more efficient end-to-end solution for operators to upload and manage applications and services.
“Operators are starting to look at ASP hosted services and not just services integrated into their own data centres,” says Wugofski. “This will facilitate a better time to market and greater flexibility. We are trying to help operators drive more mobile data revenues by providing them with a much easier development platform to work with.”
This approach goes down well with Steve Glagow, director of Orange’s Partner Program. “We can’t dictate to users what devices they should use. We have to have a broad range of handsets and applications available. The user experience is our primary concern with time to market coming a close second. For us, anything that can help us meet these two criteria is welcome news.”
The dilemma that faces operators will not disappear over night, however. Although brand and customisation issues continue to take up their attention, many users are becoming more and more sophisticated consumers of mobile data services. Dean Bubley of Disruptive Analysis, says that no matter how much the operators will try to control what kind of applications subscribers access, the user will ultimately determine this for themselves.
“Solution providers, such as Action Engine, have found an interesting way of taking Internet-based applications and content over to the mobile phone and they can do customisation services for the operators. But over time, users will become less concerned about handset customisation because a growing number of people have become used to using PC-based applications and know how to find them. They will not want operator-led branded content and applications on their phones,” Bubley says.