A Markov-Graph Cache Replacement Policy for mobile environment

The Location Based Services (LBS) have changed the way mobile applications access and manage Mobile Database System (MDS). In LBS data is moved closer to applications to increase the effectiveness and independence. This trend leads to many interesting research problems in mobile database caching. In MDS caching is effective way to improve the performance. The desired caching can be achieved by convincingly accurate prediction of data items for the present and future query processing. It is important to take into consideration the location and movement direction of mobile clients while performing cache replacement. In this paper we propose a Markov Graph Cache Replacement Policy (MGCRP) for MDS. The MGCRP predicts the future client location by first/second order Markov Model. The access patterns generated using biclustering the database and will be prefetched to reduce the network traffic and user latencies. An expressive data structure, graph is used to find the location which is not in close proximity for cache replacement. If data related to Markov-Graph location is not part of cache then replacement is based on valid scope, access frequency, data distance and data size of data item. Simulation results for cache size and query interval show that the MGCRP performs significantly better than the existing Least Recent Used (LRU), Furthest Away Replacement (FAR), Prioritized Predicted Region based Cache Replacement Policy (PPRRP) and Markov Model based Cache Replacement Policy (MMCRP).