TCP网络的拥塞预测控制策略研究

针对Internet网络系统的拥塞控制问题,提出了一种基于预测函数的拥塞控制策略。在离散化传输控制协议(TCP)动态拥塞窗口模型基础上,将IP网络转化为具有约束的预测控制,采用预测函数算法策略进行优化求解,使队列快速平稳地到达目标值。仿真结果表明该方法能够适应复杂的网络环境,控制品质优于RED算法和PID算法,具有较好的稳定性和鲁棒性,有效减少了网络不确定性带来的不利影响。

Network congestion control for TCP networks based on predictive algorithm

A predictive function congestion control strategy is proposed for the network congestion control problem present in Internet network systems. Based on the dynamic transmission control protocol (TCP) congestion model, translate the IP network into a binding predictive control question, adopt predictive function algorithm strategy for solving optimization problem that reach the target queue quick and smooth. The simulation results show that the method can adapt to the complex network environment, quality control and has better stability and robustness than PID algorithm and RED algorithm, effectively reducing the uncertainty of the network adverse effects.