快速Max-Min公平带宽分配算法稳定性分析

Stability Analysis of Fast Max-Min Fair Bandwidth Allocation Algorithm

高QoS的网络传输要以一个稳定的连接作为保证,这就要求资源分配算法要有良好的动态稳定性,但公平分配算法的稳定性还没有得到人们的普遍重视。鉴于Max-Min公平带宽分配方案的实现算法─渐次填充算法的收敛速度太慢,本文提出了快速Max-Min公平分配算法,该算法用多次平均值来快速逼近公平资源值,巧妙的回避了渐次填充算法步长和收敛速度间的矛盾,大大提高了收敛速度。最后,本文对该算法进行了摄动分析,并证明了该算法的稳定性。

Transmission with high QoS in networks depends on a stable connection, which requires an excellent dynamic stability in the resource allocation algorithm. However, the stability of the fair allocation algorithm is always understated or even ignored. Whereas the gradual filling algorithm in Max-Min fair bandwidth allocation is slow converged, and a fast Max-Min fair allocation algorithm, in this paper, is designed. In this algorithm, an iterative average value is used to quickly approach to the fair resource value. In this way, the conflict between the step-length of the gradual filling algorithm and the speed of convergence is avoided. The result is that it converges much more quickly. Perturbation analysis proves the stability of this algorithm.