Dynamic Web Pre-fetching Technique for Latency Reduction

This paper reports a dynamic pre-fetching technique in which Web caching and pre-fetching techniques are integrated. In this technique, number of subsequent links to be pre-fetched depends on the user´s interest in accessing the documents, cache contents, current bandwidth usage and maximum capacity of the existing network. Preference lists are used for maintaining user´s preferences. A hash table is used for storing the list of accessed URLs and its weight information. Intelligent agent monitors the bandwidth usage and helps the prediction engine to decide the number of Web pages to be pre-fetched. The simulation result shows that dynamic pre-fetching technique provides better utilization of bandwidth and reduces latency. Using dynamic pre-fetching technique, cache hit ratio is increased to 40%-75% and latency is reduced to 20%-63%.