8.1 Introduction

With such a diverse client base at present (which is expected to increase in the future), Internet service providers must consider how to create suitable content and a seamless service to the customer's preferred device. Present methods for determining client characteristics are ill defined and range from proprietary solutions to device-inferencing techniques. This is presently acceptable because, in part, most users' Internet experience is via the desktop computer, which has been the focus of most Web site and application design, and because the use of alternative devices such as personal digital assistants (PDA) and WAP-enabled mobile telephones has been limited. With predictions that mobile Internet access will exceed fixed line (desktop) during 2003, the demand for device-specific tailored content will increase and Web sites that can deliver such content will be the focus of much more Internet traffic.

The demands placed by this new generation of devices means that present solutions for determining client characteristics will be inadequate. Existing techniques, such as proprietary software (e.g. the AvantGo [1] browser) or inferring from information passed to the Web server (e.g. user-agent information included in the HTTP [2] packet header), are acceptable but not ideal. As the term implies, not all devices will support proprietary software and inferring characteristics from HTTP header information is problematic when devices with similar software have different characteristics or support for different media types. A better solution is to have the device characteristics passed to the server...