Research On Web Prefetching Model Based On Double Dependency Graph

Presently caching and prefetching techniques are the primary solutions used to reduce Web access latency. Web caching technique has been widely used in different places of Internet. But as dynamic documents and personal services increase all over the world, the performance of caching deteriorates significantly. As a result, Web prefetching which is a efficient way of making up for Web caching, and the most effective method to break the upper bound of caching performance——is becoming a hotspot in Web speedup research area.This paper first introduces the development of the Internet; gives the problems Internet faced and corresponding solutions; and describes the concept, classification and structure of prefetching; then summarizes existing prediction algorithms. Then the performance of page is analyzed, starting from the response time which is modeled. From that it concludes the relation between response time and the number of embedded objects is positive correlation. The performance of prefetch model is evaluated through a variety of metrics. From the different point of view, performance key metrics are classified, unified definition and description from which choose appropriate metrics to evaluate Web prefetch model.By studying the classical prediction model, this paper improves the DDG (double dependency graph) which has the better performance. Improved model first distinguishes container objects(HTMLs) and embedded objects(e.g., images, PDF) and then uses the exponential algorithm to modify the model. Experimental results show that compared with the existing prediction model, improved algorithm has better performances in lantency ratio of the page, precision, recall and traffic overhead; Especially for the page with more embedded objects, performance is more prominent.Finally, it discusses prefetch control in server side. According to network traffic, it is adaptive to change the threshold to increase or decrease the number of prediction set. Experimental results show the performance of system applying the adaptive mechanism is better but fewer peaks of traffic increase thanks to the control mechanism thus helping to reduce the negative effect of prefetching and improving the QoS.