Research Of Video Caching Algorithm Based On WSP Collaborative Caching Methodology

The rapidly development of Video on Demand service has bringing increasingly Internet traffic. Globally, the traffic caused by video has accounted for57percent of the Internet traffic in2012. With the remarkable development of wireless technology, the bandwidth resource crisis is aggravated by the numerous3G and4G users’ demand of VoD service. For the sake of meeting the increasingly demand of VoD service, the Wireless Service Providers (WSP) begin to deploy Cache server on the Mobile Switch Center (MSC) to improve user experience and decrease the backbone network traffic. Researching WSPs’cache strategies under this scenario can not only improve the hit rate, but also have significant to alleviate the server load, decrease the backbone network traffic and improve the quality of service.In this paper, we firstly analyze the basic statistical characteristic of mobile user of Sina Video, and research their viewing behavior. The research of user behavior is not only can help to mobile user demand and user viewing pattern, but also provide significant advice to WSPs’cache strategies. We find that the top10percent videos cause80percent Internet traffic by analyzing user viewing behavior from many perspectives, including user daily viewing pattern, user activity, video popularity distribution and Video rank stability. What’s more, we find the top ranked videos, such as top ten videos have high rank stability, even though the whole rank stability of mobile videos is low. All of these findings indicate that caching popular videos can restrict video’s traffic inside the WSP, decrease backbone network traffic, reduce WSP’s costs and enhance user experience by deploying cache on MSC.Secondly, we propose a baseline methodology that every MSC caches the most popular videos locally. Then, we propose the video copy caching strategy based on WSP collaborative caching algorithm. Therefore, the goal of our work is to find an optimal caching strategy which minimizes the total network consumption of WSPs. We use an Integer (0-1) programming formulation to solve the caching strategy problem. And we assume two scenarios:The first one:WSPs do not operate with other WSPs, but MSCs which belonging to the same WSP can deploy collaborative caching; the second one:the collaborative caching can be deployed between WSPs and MSCs. Finally, the optimal solution is got by using B-B (Branch and Bound Method) algorithm. Contrast to the baseline method, the scenario one’s strategy can reduce WSPs’77.17%costs, and scenario two’s strategy proves that it is better than scenario one’s strategy by experimenting with several interconnection costs between different WSPs.At the end of this paper, we propose a heuristic algorithm which is easy to implement due to the fact that the Integer (0-1) programming formulation is NP-hard and it costs us3hours and13minutes to solve an optimal solution of the dataset which contains36MSCs. And we further improve our heuristic algorithm by proposing two plans and make it has great advantage in time consumption when compares to optimal solution. What’s more, the heuristic algorithm can reduce68.28%costs for WSPs than baseline method. And at the same time, the heuristic algorithm is far faster than optimal solution and its performance near to optimal solution. Therefore, the heuristic algorithm can be implemented in the real world.