Intelligent prefetch in WWW using client behavior characterization

Improving the latency perceived by a Web client has drawn a lot of attention right from the genesis of the World Wide Web. Hierarchical caching is found to improve the latency in addition to reducing the Web traffic in the Internet backbone. Recently, Uniform Resource Locators (URL) have been found to be dynamic, i.e. the contents of the URLs keep changing every few minutes or during every visit. This proliferation of dynamic URLs renders static caching inefficient. We present a prefetching technique by characterizing the Web client browsing behavior and show an improvement in the Web client cache hit ratio in the presence of dynamic URLs. The features we considered are the trend in visit counts and the order of visits of URLs by the Web client. The proposed technique shows an average 13% client cache hit ratio even when all the visited URLs are dynamic.