An improved cache strategy based on WSP cooperation

The rapid development of Video On Demand (VoD) service especially in wireless network incurs huge bandwidth costs. To reduce the bandwidth pressure and improve the user experience, Wireless Service Providers (WSP) began deploy cache in the Mobile Switching Center (MSC) following the power law distribution observation of video popularity for the whole system. However, based on the Sina Video dataset we find that for each MSC the popularity distribution does not necessarily follow the power law distribution and hence it is beneficial for the MSCs to cooperate in cache based on the aggregated power law distribution. In this paper, we propose two models for cooperative cache among MSCs. One is cooperative cache model of MSCs of the same WSP, the other one is cooperative cache model of MSCs of different WSPs. Then we propose a novel heuristic cache algorithm for the MSCs cooperation of the same WSP by considering the aggregated popularity distribution, the individual popularity distribution of each MSC and the physical distance among MSCs. We evaluate the performance of the proposed models and heuristic cache algorithm with the previously mentioned dataset. Results show that our heuristic cache algorithm is effective and efficient the performance gain reaches 88.48% of the performance gain of optimal solution at most. And the time cost reduces to 0.003% of that in the optimal solution by using cluster. In addition, we propose a number of visits interval method to select a proper cycle of cache update to obtain higher and more stable hit rate of cache.