Abstract
Caching frequently accessed data items on the mobile client is an effective technique to improve the system performance in mobile environment. Proper choice of cache replacement technique to find a suitable subset of items for eviction from cache is very important because of limited cache size. Available policies do not take into account the movement patterns of the client. In this paper, we propose a new cache replacement policy for location dependent data in mobile environment. The proposed policy uses a predicted region based cost function to select an item for eviction from cache. The policy selects the predicted region based on client’s movement and uses it to calculate the data distance of an item. This makes the policy adaptive to client’s movement pattern unlike earlier policies that consider the directional / non-directional data distance only. We call our policy the Prioritized Predicted Region based Cache Replacement Policy (PPRRP). Simulation results show that the proposed policy significantly improves the system performance in comparison to previous schemes in terms of cache hit ratio.
Highlights
Recent advances in wireless technology have ushered the new paradigm of mobile computing
We presented a cache replacement policy, Prioritized Predicted Region based Cache Replacement Policy (PPRRP), for location-dependent data that uses predicted region based cost function for selecting data items to be replaced from the cache
In order to decide which data items to replace from cache, an attempt must be made to predict what items will be accessed in the future
Summary
Recent advances in wireless technology have ushered the new paradigm of mobile computing. While the Manhattan based policy accounts for the distance between clients and data objects, the major limitation of this approach is that it ignores the temporal access locality of mobile clients and the direction of client movement while making cache replacement decisions. Existing cache replacement policies only consider the data distance (directional/undirectional) but not the distance based on the predicted region or area where the client can be in near future. Very few of these policies [5][7] account for the location and movement of mobile clients.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have