A Data Mining Approach for Cache Replacement in Location-Dependent Environment

摘要

With the extensive popularity of mobile devices, a huge amount of research effort has been attempted towards exploring several aspects of mobile computing. In the light of many applications such as traffic conditions, stock information, weather reports and most importantly location dependent services - that require processing of location dependent queries - this area of research has gained much importance. A mobile server is expected to concurrently serve many clients. Periodic broadcasting is an important method for data dissemination in a mobile environment and is also independent of the number of users. Caching frequently accessed data items in location dependent environment for making adaptive updates, is a better method to exploit broadcasting strategy. The location dependent aspect along with the temporal aspect of the data, if applied intelligently in caching the data, is expected to give a better performance for accessing data and improving cache consistency. In this paper we intend to give, a method of predictive prefetching of data, obtained after processing the past references, to derive the probability of future access based on data mining techniques, that is expected to improve the performance of data retrieval in location-dependent environment thus providing efficient mobility management. This paper gives a data mining approach for performing cache replacement that gives accurate predictions and can also improve the system performance.