Abstract:
Video streaming has been the most popular applications in social networks. However, the current social networks still use the traditional client/server (C/S) architecture...Show MoreMetadata
Abstract:
Video streaming has been the most popular applications in social networks. However, the current social networks still use the traditional client/server (C/S) architecture, which seriously limits the scale and scalability of video streaming system in social networks. To meet the requirement of more users in social networks, more money are invested to purchase server bandwidth and storage space. Recently, P2P streaming has been a promising way to deliver video streaming and reduce the pressure of the video server. Unfortunately, the traditional mechanisms of P2P streaming systems are suboptimal, even not suitable for social networks. In this paper, we propose a P2P-assisted video streaming system in social networks, called SocialStreaming. The main contribution of this paper is: i) we put forward a network coding based storage strategy, which can improve the utilization of each peer's limited storage space. Moreover, with the same storage space, it can increase the diversity of stored data, which can improve the efficiency of data sharing; ii) we propose a social network based streaming predelivery algorithm in distributed way, derived from the Metropolis-Hastings algorithm. This algorithm could effectively deliver the video streaming to some peers, which will watch this video later with high probability; iii) we present a social network based neighbor selection algorithm, which combines the network performance factor and social factor as a tradeoff to select proper neighbors to exchange data. The trace driven based simulation results also demonstrate the effectiveness and efficiency of SocialStreaming, which can bring good transmission performance and significantly reduce the load of the server.
Date of Conference: 07-10 July 2013
Date Added to IEEE Xplore: 06 March 2014
Electronic ISBN:978-1-4799-3755-4
Print ISSN: 1530-1346