Research On Key Technologies Of Peer-to-Peer Streaming Media Distribution

The streaming media application has become more and more popular with the rapid development of Internet and broadband networks. Compared with traditional framework of client/server (C/S), peer-to-peer (P2P) technologies provide effective means of deploying large-scale streaming media distribution with low cost, high scalability as well as good quality of service and experience. Driven by application requirements, the key technologies of P2P streaming media distribution draw more and more attentions from both industry and academia. However, there are still full of challenges for providing reliable and manageable P2P streaming service, for the nature of P2P networks (free, open and centreless) and streaming service (large bandwidth consumption, real-time requirement and long transmission duration). The key technologies for P2P streaming media distribution includes overlay networks construction coupled with peer organization, management of "content" and "bandwidth" resources, coding technologies, together with mechanisms of security and management etc.. In this paper, these issues are studied, with a special focus on data storage and transmission using recoded rateless coding.Firstly, using either C/S framework or its evolution form CDN to deploy large-scale streaming media service not only leads to the system's lack of scalability and concurrent processing capacity, but also brings high cost. Taking advantage of P2P, a novel manageable P2P streaming media distribution framework using coding technology is proposed. Learning from the idea of network coding, the recorded rateless coding is processed at both the server and peer to improve the performance of data storage and transmission. At the same time, in order to maintain the index with edge servers, a "multi-domain divide and conquer" architecture is designed, attached by the entire layered function module partition.Secondly, the high dynamics and interactivity of P2P video-on-demand (VoD) system make it more complex to design and implement than P2P file sharing and live systems. Building P2P VoD system with high quality of service and experience depends on three core aspects:peer organization, data storage and data transmission. Hybrid peer organization mechanism, recoded rateless coding for data storage and transmission, as well as cross-sharing multi-channel architecture are adopted for designing a novel VoD system, namely xVoD, which has high resilience and scalability, low communication overhead etc.. In xVoD, "dynamic starting point" is introduced for neighbor nodes selection, which supports VCR operations well; cross-sharing multi-channel architecture aims to make full use of the peer's uploading and storage capacities; following "rare first" strategy, a heuristic storage allocation algorithm is used to make the distribution of each data block more uniform, which can improve the probability of getting blocks from neighbors (not server); rateless coding-based request-driven scheduling not only eliminates the need for reconciling packets inside each block, but also simplifies the retrieval of different providers; according to the request priority relying on urgency and rarity, block request is responded and bandwidth is allocated using greedy algorithm by the provider. The experiment results show that xVoD substantially reduces server stress, and also has significantly good performance in terms of startup latency, VCR latency and block seeking overhead. Meanwhile the performance improves with more storage of each peer.Thirdly, the distribution state of "content" changes constantly, according to some system parameters, the user's online status and viewing behavior, which has some influence upon playback quality and server stress. Considering the nature of streaming media application, from the perspective of resource sharing among peers, data-get-rate of P2P streaming system is defined, and the probability description is also presented. Then for the given model under some assumptions of a rateless coding-based P2P VoD system, the data-get-rate is analyzed and calculated. In addition, the modeling analysis of user behavior is given using discrete Markov chain model. By means of experiments, the relationship between data-get-rate and system parameters (including number of file blocks, number of coded pieces, number of neighbor nodes) as well as the law of user behavior is studied. Furthermore, some preliminary guidelines for VoD system design and optimization are given.Fourthly, reasonable and efficient security mechanism is needed for the operation of P2P streaming systems. It requires not only protecting against attacks from the outside to guarantee that only registered user can access services with data confidentiality, integrity and authenticity, but also detecting the data pollution from the inside to reduce the propagation of bad encoded packets. ECC-based self-certified public key cryptosystems (ECCSCPKC) is introduced to construct the system security infrastructure; and its related security schemes, with a few modifications, are used for designing an efficient authentication and key agreement (AKA) protocol. Meanwhile, homomorphic hash function is used for checking the validity of an encoded packet without decoding; in addition, some strategies are established to support effective management of the system.