Research On Media Content Distribution And Management In Peer To Peer Based Live Streaming Systems

The traditional streaming systems are based on server/client model, which is difficult in getting high scalability and brings high costs. Peer to peer based streaming systems avoid these problems by taking advantage of peer to peer technology. However, peer to peer based streaming technology faces many challenges such as high churns of the topology. An efficient and stable streaming strategy and QoS (Quality of Service) control method are the key points of such systems.Many Peer to Peer based streaming models have been proposed, as they are dedicated to resolve the different problems, and have many advantages in bandwidth efficiency, delay, and avaibability. Peer to peer based live streaming technology is an important research branch of peer to peer based systems. At present, Tree-based application level multicast and gossip-based unstructured framework become more and more popular for their promising future.In traditional tree-based application layer multicast, the leaving or crash behavior of nodes in the upper layers often cause buffer underflow. To decrease the influence,"adaptive to position"data distribution and management strategy is proposed and used in Aysee1.0, which is simple, highly efficient, and scalable. Even users can join the system and enjoy the media content at any time, deployment experience still shows the strategy can perform well.In mesh-based peer to peer live streaming system, since each peer has multiple connections, how to increase sparse media data costing least time is the key that guarantees nodes get the media data at the least time. The strategy, called UQMM which abides"fetches urgent data firstly, and amplify sparse segment"tries to resolve this problem. Through analysis and extensive experience, by the effort of peers to amplify sparse segments, other peers can get those more easily; the"trend-oriented bandwidth-adaptive"method can adjust the request quantity sent to each source peer. The algorithm does not accurately get the bandwidth value between two peers or among peers. However, it sends requests which can be adaptive (tread on the heels of the bandwidth trend) and greedy (take full advantage of available bandwidth).