Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)

First Advisor

Edward F. Watson


The World-Wide-Web has experienced exponential growth in recent years. This growth has created a tremendous increase in network and server loads that have subsequently adversely affected user-response times. Among many viable and available approaches to reducing user-response time, Web caching appears to be one approach that has recently received considerable attention. In this dissertation we explore a new approach to the study of Web cache policies, namely model-driven simulation. We present a good model of Web-access user patterns based on sound theory and principles from the information sciences. This model is justified by the empirical web access data from several different web sites. The importance of removal policies in improving cache performance motivates us to propose a dynamic and robust removal policy which incorporates the characteristics of user access patterns. We show that our proposed removal policy performs rigorously well over a variety of parameters. In this research we take a model-driven simulation approach to evaluate the impact of different factors and policies on cache performance. The results indicate that cache size, user access patterns and removal policy are major factors affecting cache performance. Continuous removal method is a good and simple method. The increase of average document size, low comfort level (less than 50% cache size) and threshold policy would degrade web cache performance. Finally, we discuss the limitations of our current research and give some directions of future research.