An analytical study of object relocation strategies for wireless environments

Caching is a commonly used technique for reducing access latency and improving scalability. However, the static nature of existing network caching techniques makes them unsuitable for wireless environments. As mobile clients move from one location to another, the performance of these caches deteriorates. To combat this problem, object relocation strategies can be used, where objects are dynamically relocated to locations near the moving clients. Existing work on object relocation have focused on achieving relocation transparency. Little attention has been given to the network overhead introduced by the relocation. In this paper, we propose a low overhead object relocation strategy suitable for wireless environments. Object lists are passed between nodes prior to relocation to ensure only the nearest copy of each object is relocated. We have developed detailed analytical models of the proposed strategy and a number of other strategies to facilitate comparison. Analytical and simulation results show the proposed strategy effectively reduce the effect of mobility on the performance of network caches. The relocation overhead of the proposed method is also significantly lower compared to existing schemes