Research On Model, Monitor And Optimization Of Mesh Overlay Based IPTV Live Streaming

With the rapid development of the Internet, the media streaming technology has achieved great progress. At the same time, increasing user number and video data have offered a new changelle for the distribution of video in IP network in large scale. As the commercial streaming service, the IPTV has to deliver multiple programs with high definitions to a lot of people at the same time, and all of these have more requirements for the technology of service fulfillment, assurance, and billing. All of the current technologies can't satify them. The emergence of Peer-to-Peer(P2P) technology based IPTV, which has low cost and good scalability, utilizes the network bandwidth and the capacity of user-end systems which were ignored in the past become the main direction in the IPTV delivery research. Among all the related P2P architectures, the Mesh with high scalability, reliability and better performance in a dynamic Peer-to-Peer network environment, is a hot research topic in recent years. But there is still a lot of work to do to really meet the requirements of IPTV live streaming service.Based on a systematical summary of related works and future direction on IPTV delivery technology, this dissertation focuses on related technologies of Mesh overlay based IPTV live streaming and gains several achievements on some sub-topics. The major contributions of this dissertation are as below:1. This dissertation proposed a production model of video chunk within its whole lifecycle in Mesh overlay based IPTV live streaming. The production is the number of peers who own video chunk m at time t and represents the maximum supply capability of special chunk in theory. We extend the model from unlimited buffer to limited buffer, and extend the analysis from born, delivery to hold and disappear of video chunk, split the production function into the growing phase, the stable phase and the fading phase, study them seperately. We discuss the influence of various key parameters of the system to the production and conduct the concrete formulas of production function under typical topologies. The simulation results reveal that the model and analysis is consistent with the reality, and show the production function in various and complex enviroments.2. The dissertation presents an unbiased random walk based service monitoing algorithm, which can overcome the overlay dynamic and large scalability, service quality changing, and is fit for the Mesh overlay based IPTV live streaming. We discuss the feasibility of unbiased random walk based sampling algorithm in Mesh streaming overlay from theory, its influence factors, its performance, and finally verify the algorithm by simulation and the results show that it is accurate, real-time, efficient, and also has low control costs, which is fit for the enviroment. We also show the impact of different setting. The algorithm can overcome the shortcomings of current methods in some aspects.3. The dissertation proposes a priority preemption algorithm (PPA) which is used for the delay optimization in the Mesh overlay based IPTV live streaming. The method makes use of the heterogeneity of nodes in bandwidth and other aspects, and to overcome the difficulties which are brought by Mesh architecture. We identify the main distribution architectures dynamically and periodly, and make the main distribution architecture to relay preemption message. Based on the neighbor relationship model, the algorithm make the high capability nodes to stay in the high level of the main distribution architectures to reduce the delay, and improve the system overall performance, and also the service quality of themselves. The simulation results show the PPA can reduce the delay without quality degrading and high cost.4. The dissertation proposes an integrated service supporting architecture which overcomes the shortcomings of pure P2P and CDN, and is fit for the mesh overlay based IPTV live streaming. The architecture can monitor the service quality, and improve them according to the needs. Its componet, supporting nodes, initialize parallel random walks to monitor the service quality and analyse them. When the service quality of some channel is below the threshold, the architecture will add the upload bandwidth of the supporting nodes to fit for the workload changing. The supporting architecture will also allocate bandwidths between different channels, which mean different overlays, to make the revenues as more as possible. We finally verify the feasibility, effectiveness and scalability of the proposed architecture by simulation.