Research On Congestion Control And Peer Selection Optimization Of Peer-to-Peer Video-on-demand

Video on demand(VOD)is an interactive system that can play and transmit videos.It can meet users' requirements that can play and watch at their will.The combination of P2 P and VOD technology can take advantage of computing and storage capacity of distributed peers to relieve pressure on servers.However,there are some problems such as unba lanced network load and peer selection failure resulting in longer video buffer waiting time.This leads to poor users' experience.Therefore,the research of network load balancing strategy and peer selection method can effectively reduce the waiting time of user video on demand,and ensure the fluency of video playback.The main work of this paper is as follows:1.Related theories and technologies are introduced.Firstly,the P2 P technology and classification are introduced,and the P2 P model are compared with C/S model.At the same time,we expound the classifications of P2 P network model and the characteristics of each classification.Secondly,this paper introduces the concept and classification of load balancing technology.2.Congestion control based load balancing routing in unstructured P2 P networks is proposed.In order to solve the problem of network load imbalance caused by peers' congestion in P2 P networks,This paper proposes a new churn-resilient protocol to assure alternating routing path to balance queries among peers under network churn.Our proposed protocol uses two strategies to make queries balancing among inter-and intra-group peers.On the one hand,a resource grouping and rewiring strategy is proposed to periodically cluster peers having same set of resources.This strategy makes locating resources more efficient in the inter-group peers and balances the query load among different source groups.O n the other hand,load balancing routing is achieved by collaborative Q-learning among peers.By using the collaborative Q-learning method,queries are guided to avoid forwarding to those congested peers.Thus,query routings are forwarded balanced among intra-group peers.O ur simulation results show that the desired resources are located more q uickly and queries in the whole network balanced.O ur simulation results show that our proposed strategy makes queries among inter-group or intra-group peers avoid forwarding to those congested peers.Thus,the network load is balanced.3.The optimal selection strategy of P2 P VOD based on linear programming and reinforcement learning is proposed.In order to select appropriate peers in P2 P VOD system to make full use of source in P2 P network and reduce the server bandwidth consumption,the optimal selection strategy of P2 P VOD based on linear programming and reinforcement learning is proposed.The strategy is divided into two tiers.In the first tier of the optimal peer selection strategy,servers exploit the global information collected periodically from the peers to guide the peer selection process.The second tier of the optimal peer selection strategy uses the collaborative Q-learning method which learns from local information of neighbor peers such as remain uplink bandwidth,number of resources,and peers' state of congestion to find the peers capable of availing the required bandwidth to the request peers in a fully distributed fashion.