In this dissertation, we address the problem of adapting Web content to improve access by wireless devices and other resource limited clients. We present a Web middleware architecture that supports the customization of Web content and streams, called the Web Stream Customizer Architecture (WSCA). Our thesis is that the WSCA improves upon proxy-based customization of Web content by using intermediaries called Customizers to support flexible, adaptive customization, yet by using existing Web mechanisms and focusing on HTTP only, it remains deployable and avoids the complexity of more general customization systems. Customizers intercept and act on HTTP stream to improve performance, reliability and security, supporting the following customization functionality: (1) Stream buffering; (2) Traffic regulation; (3) Filtering or modifying the data as it passes through; (4) Repeating, redirecting, and generating HTTP requests; (5) Storing data for remote access.; This allows such functionality as image filtering, storing transaction results in case of connection failures, encrypting sensitive data, and buffering streaming video near the client to improve playback quality, all of which are especially valuable for wireless clients, which are our primary target of support. We introduce the use of a personal server, which supports running customization modules and maintaining user configuration and application data, while third party servers can be used to bolster performance, increase fault tolerance, and satisfy special security requirements. Designing the system specifically for the Web allows us to make a number of simplifying assumptions regarding the programming model, the user model, and the system design and implementation. Our architecture is more flexible than existing static and dynamic proxy systems, and less complex than more general mobile code systems, and is deployable, while incurring low overhead. We present experiments showing that the benefits of using Customizers far outweigh the overhead of using them, and analyze the performance of a number of Customizer applications, including an adaptive image filter and a video buffer that improves the performance of streaming video.
展开▼
