Research Of Video Sharing Technologies Based On Social Network

With the development of P2 P streaming system, it tends to generate a tremendous amount of inter-ISP traffic for establishing a large number of random inter-ISP connections. Therefore the optimization algorithms to reduce inter-domain traffic is needed to be studied. In this context, we present an inter-domain traffic optimization algorithm based on the localization strategy. Furthermore, with the rise and deployment of social networks, it showed some good characteristics such as the aggregation based on friendships, preferences and interests. Thus, we introduce social networks into P2 P streaming systems and build the overlay based on social network to optimize video prefetching and caching strategies.In this thesis, we firstly investigated the effects of inter-ISP connections on the streaming quality and modeled the inter-ISP traffic problems. Then we researched the existing inter-ISP traffic optimization algorithm and proposed an ISP-friendly three-tier overlay(HOPES) to reduce the inter-ISP traffic based on localization. HOPES construct an overlay composed by upper, middle and bottom overlay. In HOPES, we decrease the inter-ISP traffic by limiting the number of inter-ISP connections, while limiting the depth of inter-ISP path to minimize the delivery delay. In addition, we adopted emergency chunk scheduling and retransmission mechanism to reduce the chunk loss rate. We performed the simulations in various scenarios to evaluate the performance of HOLES, and the extensive simulation results showed that HOLES can greatly reduce inter-ISP traffic with minimal impact on the streaming quality.We studied the video-on-demand(VOD) system with the social network and then proposed some algorithms based on the characteristics of social networks. In this thesis, we proposed an overlay construction algorithm based on social network, which uses relationships and user interests to optimize the overlay performance. In this algorithm, a real overlay is constructed by friend peers and ordinary peers, while a virtual friend network is built by virtual connections between friends.In addition, we designed a social network-based prefetching and caching strategy to improve system performance and user experience. In prefetching strategy, we firstly calculate the similarity of videos watched by user neighborhoods and increase the similarity of videos watched by friend peers. For cache management, we first define the buffer value function as the measure of the video whether to be cached. We calculate the value of videos based on the popularity, cache space and friend relationships, and then cache the video according to the value.Finally, in order to verify the effectiveness of these algorithms, we made experiments under a variety of environments using two simulator-LiveSim simulating live streaming systems and P2 PVODSim simulating video-on-demand systems. The simulation results on HOPES showed that the algorithm can effectively reduce the inter-ISP traffic and to improve video playback quality. Then we simulated the algorithms based on social network and the results showed the effectiveness and feasibility of these algorithms.