对等网端到端多路径选择建模及算法研究*

在对等网上利用多路径分发视频是一种重要的机制,虽然在一对节点之间找出符合条件的多条路径并不困难,但发送端如何从可用路径集中选出最优路径子集,并为其最优地分配发送速率仍是一个难题。为此,提出一种新的对等网端到端最优多路径选择与速率分配(OMPSRA)算法。首先,应用排队论建立OMPSRA模型,并推导出一种新的OMPSRA公式,公式既给出最优分配的计算方法,也给出路径的最优速率分配与各路径最大可用带宽之间的关系,利用此关系可选出最优路径子集。最后基于公式实现OMPSRA算法。理论分析和仿真实验结果表明提出的算法能对通信量进行全局最优分配,最小化视频传输的端到端时延,有效提高视频传输质量,比同类算法有更好的性能。

Research on Modeling and Algorithm of End-to-End Multiple Path Selection in P2P

The multi-path transport is an important mechanism for supporting video distribution in P2P Streaming .While algorithms for finding a set of paths between a source and destination are not difficult to develop, it remains a challenging task to select an optimal subset of paths and optimally allocate streaming rate among multiple paths of selected subset to transport video and maximize its quality. To cope with the problem, a novel solution (OMPSRA algorithm ) is proposed. Firstly,the problem is modeled as a non-linear optimization problem by applying queueing theory. And then, the OMPSRA formula is derived ,which gives the method of computing the optimal solution and selecting an optimal subset of paths. Finally, the OMPSRA algorithm based on the formula is proposed. The theoretical analysis and extensive simulation results using NS2 shows that the proposed OMPSRA algorithm can minimize video transmission delay , optimally allocate traffic among multiple paths of selected subset to transport video and maximize its quality , and achieve a better performance than the related works.