Exploiting Online Social Networks To Accelerate Peer-Assisted Content Distribution

With the popularity of online social networks (OSN), the P2P application built onthe social network is emerging. The unique feature of online social network is thatnodes in the online social network are associated with people in the real world.Generally, nodes with friend relationship will share common interests. And nodes tendto help their friend nodes. It can reduce the probability of free riding and improvesearch efficiency of resources. On the other hand, nodes with more friends in thenetwork are more likely to stay in the network for a long time. Thus, the OSN-basedP2P application can address the problem caused by high churn in the P2P network. Itprovides a new direction to solve the inherent incentive problem of P2P systems byexploiting the features of online social networks.BitTorrent, as a typical P2P content delivery application, introduces a “Tit-for-Tat”(TFT) mechanism to limit free riding. TFT mechanism can effectively prevent nodesfrom acting as a “free rider”. But at the same time, it also limits the distributionefficiency of the whole system. The problems caused by TFT mechanism is given asfollows:(1)It will take long time for a new node to get “the first piece” of the resourcefile. Because the new node has nothing to upload to others, according to the TFTmechanism, it will not be able to receive any upload service from others.(2)TFTmechanism will cause the deadlock of the nodes. If there are two nodes each of whomis sending the request to the other one, but either never offers upload service to theother in the last20seconds, in this case each of them will wait for other’s uploadservice instead of proactively offer the upload service to the other one.In order to solve the problem that TFT mechanism hinders nodes from more efficient cooperation, this paper proposes an adaptive chocking algorithm based ononline social network. Our main contributions are summarized as follows:(1) Theimproved algorithm introduces a value that records the amount of upload contributionbased on the reencounter possibility of social nodes, so that it can encourage nodesto actively provide upload services;(2) The social nodes regard their friend nodes astrusted nodes because they are connected based on the real world relationship.Therefore, the social node will unchoke three friend nodes with the highest socialcloseness value and one other node with the highest historical contribution value;(3)When the number of friend nodes in the social node’s friend list is less than three, theadaptive algorithm will shift the correspond “unchoked amount” to other nodes,which make full use of the node’s upload bandwidth. Finally, our simulationexperiments confirmed that our proposed algorithm can effectively reduce the averagestartup time delay and download completion time compared with the traditionalTFT-based algorithm.