This position paper describes the non-PC embedded platform for the next generation Web.
Japan is a leading wireless Internet country with 16 million cellular phone-based Internet users at the end of July. The subscriber increase is about 2.5 million during July. This explosive growth is supported by both of skilled software integration in embedded systems like cellular phones, game consoles, and PDAs and content-awareness from a very early stage of adoption. 80% of cellular-phone based Internet is based on W3C-compliant content format, to make use of full skills of the current Internet.
ACCESS is an Internet solution provider company with 17 years of experience in embedded software. We license Internet software such as browsers, TCP/IP, mailers, Java-VM, SSL modules to hardware vendors. With 60 customer's products, the total shipment reaches 13 million, especially, 1.2 million micro-browser shipment to W3C-compliant cellular phones every month. We have a flexible layered software architecture with proprietary WAVE layer to abstract underlying hardware and user interface in embedded network software.
This paper positions how general framework of multi-modal architecture is implemented in a flexible non-PC software architecture.
Even with the emerging market needs and explosive expansion of the market, the hardware capacity is limited in non-PC devices, especially in cellular phones. The memory and CPU capability is limited. The most important challenge is how we can trade off the content providers' task and embedded software task.
As described in [CompactHTML], there are several hardware restrictions in mobile devices. The major characteristics are as follows:
It is easy to expand something. On the contrary, it is very difficult to make a concise, consistent subset of something.
We should use the Internet standards as much as possible for Wireless and Mobile Internet.
Multimodal Web should be discussed under this principle.
There are two types of Multimodal Web.
In the second case, the API among browser interface will be one of the key issues.
With the multi-modal web, the relations between various modal interfaces will be more complicated. The user interface's modality will be very context-aware. Therefore, it should be very modular and flexible to be selected by a end-user or a content provider to choose the best possible modal interfaces. With the multi-modal web, the software architecture should be very flexible and modular to facilitate the feedback from the content providers. In addition, the development environment which can be usable in many PC and non-PC devices will be necessary. Multimodal Web will explore rich possibility in non-PC devices in many new contexts.