Abstract
Digital images have become a dominant information source on the Web. Emerging Web-enabled applications, such as global collaboration in environmental studies, point and click manufacturing, and electronic commerce, to name a few, push for timely processing and transmission of images. This real-time imaging is much more than variations on image processing without regard to time. Its performance requirement on networking is beyond the capability of the current Internet, which provides a best-effort service model compromised with end-system traffic management. For the Web to support real-time images, we need to understand real-time image features, design a new Web service model, and convert a user's application requirement into a system service agreement. This paper presents a Web architecture that provides predictable and different service levels for specific quality of service (QoS) requirements, called QoS Web. We examine the Web image features in depth, which leads to a generic formula describing a user's preferences for image quality and timing constraints. It then maps onto the QoS requirements as resource parameters expected from the Web. The influence of the QoS settings on the perceived image performance is analyzed in theory and tested with experimental simulation. We present a procedure for users to specify their requirements to the QoS Web for real-time image transfer. In addition, an inverse mapping mechanism is included for re-negotiation when there is a shortage of network resources or the user's requirements change.