一种惩罚非适应流的无状态主动队列管理算法

Stateless AQM algorithm for punishing non-adaptive flow

CHOKe是一种无状态的主动队列管理算法,能以较低的代价提高网络的性能,但是由于它对超速的非适应流的惩罚力度不够,不能够很好地实现带宽的公平分配。针对这一问题,提出一种新的对非适应流的惩罚机制-ECHOKe算法,通过比较分析FIFO队列中随机抽取m个包与新到包的流ID号之间的关系以及m个包之间的关系建立了新的数据包丢弃机制,从而实现在不提高算法复杂度的前提下增大对超速非适应流的惩罚,提高了公平性。仿真结果表明,在不保持流的状态下,该机制对非适应流具有更强的识别和控制能力,与B-CHOKe和M-CHOKe算法相比,能够进一步加强对非适应流的惩罚,实现更为公平的带宽分配。

CHOKe is a stateless AQM algorithm, which can improve the performance ot networK, out its fairness is not so good because of the slight penalty for non adaptive flow. Based on the study of the CHOKe algorithm, a new algorithm- ECHOKe to penalty for non-adaptlve flow in is proposed this paper. Through analyzing the relationship between the m packets random selected from the FIFO queue and the new arriving packet＇s flow ID and the relationship among the m selected packets themselves, the ECHOKe algorithm establishes the dropping data packet mechanism, consequently increases penalty for non-adaptive flow without enhancing the algorithm complexity, thus improve fairness. Simulation results indicate that the proposed mechanism has stronger recognition and control ability for non-adaptive flow without keeping the flow＇ s state information. The ECHOKe algorithm has better penalty performance for non-adaptive flow for fairness of bandwidth allocation than that of the B-CHOKe algorithm or the M-CHOKe.