Research On High Efficient Data Delivery Tecnology Based On Broadcast/Unicast Hybrid Architecture

With the fast development of the Internet,streaming media has become the mainstream Internet service.It is estimated by Cisco that streaming media will occupy 80%~90% of the whole Internet traffic.However,the available bandwidth under wireless condition is quite limited and the stability of wireless channels is very poor.Thus,delivering streaming media under wireless conditions efficiently is of great significance.Usually,there are two means to deliver streaming media in wireless environment.One is to broadcast the hot items which have common requirements so as to save bandwidths,and the other is to deploy caches on the edge of the network.Thus,two kinds of networks are born,namely,content delivery network(CDN)and Peer-to-Peer(P2P)network.In CDN,request redirection and load balancing techniques are adopted to release the traffic load of backbone network.In P2 P,computation and cache abilities of end devices are efficiently used via distributed architecture.Nowadays,there has been growing concern about we media.In popular we media platforms,users can upload videos without limit,so that the amount of streaming media transmitted online grows rapidly.On the other hand,users can be deeply involved in media and some controversial topics can become hot news within a short period of time.As a result,the discrepancy between limited bandwidth and growing scale of streaming media service becomes deeper and deeper.This article works on three aspects to realize the goal of delivering streaming media under wireless conditions efficiently:Firstly,the modified streaming media popularity distribution model is studied.A modified Zipf model is proposed based on analysis of hot index of streaming media news.Further,the Gamma model is adopted to predict future hot index.The proposed Zipf-Gamma model fits the real data better,thus it can reveal the characteristics of popularity distribution in streaming media contents.Secondly,the efficient broadcast scheme under restricted wireless conditions is studied.The optimal average access time of data items is gained from the dynamic channel allocation function under different load.Meanwhile,streaming media contents can be transmitted via broadcast channel or unicast channel according to their popularity.The simulation result shows the proposed algorithm can achieve minimum average access time considering the dynamic changes in contents' popularity.Thirdly,the dynamic CDN architecture is studied.CDN edge servers can reduce the traffic load greatly in wired conditions.However,the mobility of CDN servers poses huge challenge to the system's performance under wireless conditions.The content concentric streaming media transmitting scheme is proposed within dynamic CDN structure.The influence of server speed,server radius,cache scale is studied to fully understand the performance of dynamic CDN nodes.