Systematic Multiresolution and Its Application to the World-Wide Web

Many emerging environments are increasingly facing the problem where the data and computational requirements of applications easily outstrip the system resources. This is particularly acute in the World Wide Web (WWW) and many data-intensive applications like OLAP and multimedia databases. In this paper, we address this problem via multiresolution, i.e., providing responses at different qualities resolution and costs. Furthermore, we demonstrate the benefits of providing systematic multiresolution. We argue that a fully effective multiresolution system consists of more than simply generating different resolutions - it involves careful end-to-end system design and evaluation. We identify the requirements of systematic multiresolution and propose a general framework for doing so. We validate our conceptual contributions by implementing a proxy-based multiresolution system on the WWW. Our design addresses two key problems facing the web: high latencies and heterogeneity of client resources and requirements. To further enhance performance, we introduce novel multiresolution-aware caching techniques. We demonstrate the utility of multiresolution and the caching enhancements using extensive experimental studies.