多层网TCP-RED非线性流体模型的高效网络拥塞控制

Effective Network Congestion Control Based on TCP-RED Nonlinear Fluid Model for Multilayer Network

提出了一种多层网络拥塞控制方法,该方法基于流体模型的微分方程设计多层网络拥塞控制.此外,该方法利用流体模型的稳定性保证网络性能参数的均值在多层网络环境及大数据流的情况下向平衡点收敛,同时也有效限制了AQM参数的设置.理论分析和仿真结果证明了这种网络拥塞控制方法能在多层网络中避免拥塞,并当网络出现大数据流时,该多层AQM系统能充分利用网络带宽及保持路由器队列长度的相对稳定.

The senders and receivers are connected by a pair of routers. In this paper, we present a netork congestion control method in multilayer network with differential equations＇ fluid model. The proposed network congestion control method considers that the stability of the fluid model can ensure the convergence of the average rate to its equilibrium point under many flows in multilayer network, and may impose excessive restriction of our choice of AQM （Active Queue Management） parameters. Theory and simulation results verify the proposed congestion control method can avoid of the congestion problem of multilayer bottleneck network. Moreover, when the network encounters many flows with large capacity,AQM system utilization still approaches to full utilization, while the buffer size on the order of lmndwidth-delay product is necessary for the stability of the fluid model with full utilization.