Efficient web browsing with perfect anonymity using page prefetching

Anonymous web browsing is a hot topic with many potential applications for privacy reasons. The current dominant strategy to achieve anonymity is packet padding with dummy packets as cover traffic. However, this method introduces extra bandwidth cost and extra delay. Therefore, it is not practical for anonymous web browsing applications. In order to solve this problem, we propose to use the predicted web pages that users are going to access as the cover traffic rather than dummy packets. Moreover, we defined anonymity level as a metric to measure anonymity degrees, and established a mathematical model for anonymity systems, and transformed the anonymous communication problem into an optimization problem. As a result, users can find tradeoffs among anonymity level and cost. With the proposed model, we can describe and compare our proposal and the previous schemas in a theoretical style. The preliminary experiments on the real data set showed the huge potential of the proposed strategy in terms of resource saving.