The Research On The P2P Stream Media Optimization With DHT Network

In the past few years, with the popularity of the Internet, P2P streaming media systems have been widely used by a great number of Internet users. Typical P2P streaming systems have gained a large market share, such as PPLive, PPStream, etc. The fierce competition of P2P streaming systems not only promotes the technological progress of the industry, but also ensures the value of academic research. Video stream is expected to constitute86%of Internet traffic by2016according to Cisco. We can see that P2P streaming has become an area with huge number of users, and a field full of research potentiality and significance.P2P streaming system effectively avoids the server overload and weak scalability in the C/S streaming way, has good economical efficiency, and gets the attention of academic and industrial circles gradually. P2P streaming is divided into tree multicast and mesh multicast. Compared to tree multicast method, mesh multicast method based on Gossip protocol has the characteristics of high fault tolerance and high bandwidth utilization; gossip multicast type P2P streaming system has become the mainstream. Data dissemination with gossip protocol in high-dynamic environment is uncertain, the dynamic network could easily lead to a slow start and unsmooth play in streaming media system, this situation will seriously affect the user experience.How to obtain the missing data quickly and accurately became the key to solve the problem of gossip data transmission. Taking advantage of the ability that DHT network can obtain global data quickly and accurately and the heterogeneity of nodes, we organized nodes with high performance to constitute a DHT subnet for locating the missing data, improved the performance of gossip multicast. Meanwhile, we designed a full strategy for data transmission and scheduling, and verified the system both on theory and experiment.As we all know, DHT network has strict topological restriction, stable link in DHT network is always cut off by short lifetime node, causes churn, and harms the topology structure. In order to reduce the churn of the DHT network, we focused on the time features of nodes, and proposed the churn prevention strategy in DHT network based on gray theory. And we designed the middleware Churnless to achieve the strategies above.Experimental results show that the proposed method can effectively adapt to the highly dynamic network environment, improve gossip streaming playback continuity, and also speed up the startup speed of the streaming media.