Research Of Caching Algorithm In Video On Demand

With the development of network, video compression and computer storage technologies, video on demand(VoD) system has been the mainstream of current network services. But due to the increasing number of users, the network congestion, server workload and the quality of service(QoS) become the big problems. The introduction of proxy server can effectively alleviate these problems, by arranging more servers close to the users, it could not only reduce the pressure on the original server, but also improve the quality of service. However, since the storage capacity of proxy server is small, how to effectively cache video data on the proxy server has been the main problem.Since the access features of VoD systems are different from traditional Web pages, this thesis first analyzes some statistical data of video servers, summarizes several key points, and thus designs the data simulation tools. Also this thesis uses the Network Simulator(NS) build a VoD system network architecture based on proxy server.And then this thesis analyzes the traditional web caching strategy, integrating with the current video caching strategies and the characteristics of the video data, designs a new video data cache algorithm called trine lists with retention time (TrineRT). This method uses apri_list, check_list and work_list to distinguish the popularity of video data, and proposes a new concept of retention time to effectively calculate the weight of cached video data, and also proposes a new evaluation parameter Replacement Number(RN). The experimental results show that:the method could (1) effectively increase the byte hit ratio(BHT), reduce the user latency and guarantee the QoS; (2) lower the server workload by reducing the replacement on the server.