The Research On Data Scheduling And Incentive Mechanism For P2P Media Streaming

With the fast development of Internet and the unceasing enhancement of user's single processing ability, P2P media streaming not only innovates people's concepts of traditional media thoroughly but also changes their ways of work, daily life and recreation. Meanwhile, it is becoming one of the hottest topics in the field of computer science as well.As for data scheduling in P2P media streaming, starting form the current data scheduling algorithm, the widely used algorithms such as optimal media data assignment (OTS) and multi-sender data scheduling aiming at the minimum of packet loss are analyzed in detail in our research in which there is still common room for improvement——unfairness of data scheduling. In order to solve this problem, the computation method of three important elements of contribution factor, online duration and packet loss are discussed. Finally, a new algorithm model having incentive function is proposed which could dynamically adjust bandwidth allocation of peers according to their packet loss and contribution rate to realize maximum of receiving peer's interests rationally and fairness of the whole system. In the experiment, a design proposal is presented to make out the incentive function. The experimental data show that the algorithm model motivates peers to participate in data scheduling and ensure the transmission efficiency to some extents, thus realizing system fairness.In the aspect of incentive mechanism, due to the fact that the existing incentive mechanisms could not come up with unified criteria, this paper studies the formula of optimal security portfolio strategy for small traders in stock market in the framework of game theoretic analysis, discovering the similarity between peers in P2P streaming media and small investors in stock market. Therefore, a novel econometric model of P2P networks incentive mechanism is proposed by improving the formula of optimal security portfolio strategy for small investors which incentives peers to participate in uploading and significantly prevents from "free rider" and collusive behavior.Because of a variety of objective factors of peers in real networks, the simulative peers in this paper could not correspond to them exactly, and hence the work in data scheduling requires examining in networks and further improvement. In relation to econometric model of incentive mechanism, the situations turn out to be more complex and it has not been put into practice. So, there is still a lot of work to finish.