On Chunk Delivery For P2P Live Streaming Under Heterogeneous Peers

Mesh-Pull P2P live streaming has been paid much attention of research and industry field because of its high capability of coping with peer churn and utilizing peers'upload bandwidths, and high scalability. However the approach has some problems:customers in Internet usually suffer from the long startup delay; with the development of mobile communication and the evolvement of emerging network among telecoms network, computers network and broadcast network, it would be possible that mobile temernals, personal computer and set-top boxes share the streaming in Peer-to-Peer way and but the high startup delay problem also exists in this emerging scenario; to the best of our knowledge, system scale and its relationship with key factors have not been analyzed clearly, which results in arbitrary configuration of P2P live streaming key parameters so as to increase server bandwidth cost or decrease the system performance.Based on current status and development tendency of P2P streaming, we deeply explore Pull-based chunk delivery approach and system scale of P2P live streaming system for the above problems, and the contribution of this dissertation includes:1. Propose a Mesh-Pull chunk scheduling model and algorithm of P2P live streaming in the traditional single-rate scenrario. We analyze the characteristics of chunk scheduling in this scenario:uniform requirement of video quality, time characteristic of a chunk, unconstrained download bandwidth, heterogeneous peers'upload bandwidths and propagation delays. And then we propose a chunk scheduling model with local optimal delay under the guarantee of high quality video. This model provides guidance for designing a chunk scheduling algorithm attending to achieve the local minimum delay. Finally, we propose a chunk scheduling algorithm which utilizes TOPSIS to solve a problem of multi-attribute decision making like the assessment of the chunk priority in this scenario, adopts delay-optimum chunk assignment approach and an aperiodic chunk scheduling way and develops a chunk assignment algorithm by which scheduling peer can receive the maximum number of chunks from the most neighbor peers. Simulation results show that our scheduling algorithm could improve the delay performance under the guarantee of high quality video.2. Propose a Mesh-Pull chunk scheduling model and algorithm of P2P live streaming in the emerging scenario of telecommunications, computers and broadcast network. We explore the characteristics of chunk scheduling in this scenario:heterogeneous requirement of video quality, layer and time characteristics caused by cumulative layered coding, limited peers'upload bandwidths, heterogeneous peers'upload bandwidths and propagation delays. And then we propose a chunk scheduling model with local optimal delay under the guarantee of high quality video. Finally, considering the above characteristics of this scenario we propose a chunk scheduling algorithm, which adopts the TOPSIS to solve a new problem of multi-attribute decision making(the selection of candidate neighbor peer), and considers more complex attributes like layer and so on to solve the assessment of the chunk priority. Simulation results show that our algorithm could improve the delay performance under the guarantee of high quality video.3. Propose a scale model of P2P living streaming system. We analyze the factors that affect the scale of P2P live streaming system such as source server bandwidth and model of peer arrival. And then we analyze the process of chunk delivery and propose the scale model of P2P living streaming system. In accordance with this model, we quantitatively analyze the relationships between the scale and key factors such as source server bandwidth, model of peer arrival and number of substream, and propose the strategies of decreasing the provision bandwidth of source server. Simulations validate the theoretical analysis about the relationships between these factors and the scale of system so that the relationships can guide the deployment of P2P live streaming system.