Dynamic Policies in Selecting a Caching Set for a Distributed Mobile Computing Environment

The bandwidth of a wireless network is usually very limited as compared to that of a static network. Therefore, caching the right set of data into the limited storage of mobile computers is an important issue in mobile database applications. Caching not only reduces the contention on the narrow bandwidth of a wireless channel, but also improves the response time for read operations. However, we show that the problem of nding an optimal caching set so as to minimize the bandwidth utilization is NP-hard. In order for the system to determine the caching set in real time, two heuristic methods, the baseline algorithm and the greedy algorithm, are proposed. The closed-form solutions of various performance measures for the no caching policy, the baseline algorithm and the greedy algorithm are derived based on the queuing model. Experiments are also conducted for various access patterns. In general, the proposed caching algorithms outperform the no caching policy under all cases. In addition, the capacity of a cell is signiicantly improved when the caching algorithms are adopted. Our algorithms are simple and eecient, hence they can be readily incorporated into a mobile database system to ensure good response time and the cost-eeective usage of the mobile computer memory and the wireless bandwidth.