Resource Allocation Based On Auction In P2P Content Distribution Networks

P2P content distribution is a hot topic in the network application. Due to this,Bandwidth resource allocation as a key problem in P2P content distribution networks hasarisen widely attention of industrial and academic circles. Most of traditional resourceallocation schemes are usually only consider the resource utilization, user fairness andsecurity, and to ignore the non-cooperative behavior between the nodes. But in realitynetworks, there are a lot of non-cooperation nodes, which are called self-interested nodes.These nodes seize a large number of network resources, and don’t want to provide servicesto other nodes. Then,“tragedy of the commons” can’t be avoided, which seriously affectsthe quality of services for other nodes and restrict the advantage of P2P networks. Thisdissertation makes a detailed study on bandwidth resource allocation based on auctiontheory.Firstly, a simply auction-based bandwidth allocation mechanism for P2P networks isproposed. Through the effective bandwidth payment scheme, the selfish nodes have noincentives to lie and provide the real bandwidth requirements. Also the “tragedy of thecommons” can be avoided by the nodes’ healthy bandwidth competition. But when thereare a lot of strategic buys in the system, the above scheme can’t find a stable optimalsolution. To solve the problem, a novel resource allocation scheme based on VCG andKelly mechanism is proposed. It is proven that strategic buys can obtain Nash equilibriumpoint and resource allocated effectively. Finally, as the decentralized complex nodeinteractions and the double node role as resource provider and consumer amidst resourceconstrains cannot be addressed perfectly by single-sided auctions. We introduce a doubleauction market framework to solve the challenges above. Through the scheme, equalrelationship is formed between the two sides.Simulation results show that the proposed schemes not only suppresses free riding nodes efficiently, but also shortens the average completion time for content distribution anddecreases the ratio of data uploaded by the source server. Meanwhile, the utilization ofresources and the scalability of system are enhanced. So, our study has significant meaningfor promoting further development of P2P content distribution applications.