Semantic data caching strategies for location dependent data in mobile environments

Abstract

A new model for semantic caching of location dependent data in mobile environments is proposed. In the proposed model, semantic descriptions are designed to dynamically answer the nearest neighbor queries (NNQ) and range queries (RQ) from the client cache. In order to accurately answer user queries, the concept of partial objects is introduced in this paper. Both NNQ and RQ results are stored as semantic regions in the cache. We apply a cache replacement policy RAAR (Re-entry probability, Area of valid scope, Age, Rate of Access) which takes into account the spatial and temporal parameters. The cache replacement policy which was proposed for NNQ is also applied to evict the cached semantic regions. The experimental evaluations using synthetic datasets show that RAAR is effective in improving the system performance when compared to Least Recently Used (LRU) and Farthest Away Replacement (FAR) replacement policies. © 2011 Springer-Verlag Berlin Heidelberg.