A Predictive Framework for Web Access over Heterogeneous Wireless Networks

We present a framework that uses both prefetching and location prediction together to effectively lower the cost and total access time for mobile web users. A new prefetching algorithm is presented that dynamically predicts the likelihood of the mobile moving into a different network. We examine a heterogeneous system composed of two networks: a cheaper and faster WLAN network, and a slower, more expensive cellular data network. Using current industry network rates, our model demonstrates that mobility can significantly raise or lower the prefetching threshold. An accurate mobility prediction scheme can therefore achieve a lower total cost for end users.