Research On The RED Algorithm Base On A Fuzzy Smith Predictive Control System

Currently,Network Congestion Algorithms focus on three aspects：based on network source,based on network terminal and based on Cybernetics.This paper mainly studies the network congestion algorithm based on Cybernetics.The concept and the causes of network congestion are expatiated,as well as the network congestion research status at home and abroad.On the basis of analyzing of the common congestion algorithms and control model,PID Controller’s parameters of the Fuzzy Smith Controller are optimized,the large time delay error compensation of the controller in the network system is reduced,and the improved controller is applied to the RED algorithm that according to the actual situation of the network to improve the RED algorithm.The main work is as follows：(1)In order to immprove the stability of Fuzzy Smith Predictor Controller,in this paper,the Particle Swarm Optimization Algorithm is applied to the fuzzy Smith predictive controller.On the basis of analysis the degree of influence on the stability of the controller,which is caused by different values of the parameters kp,ki and kd in the internal model control structure,we combine with the characteristics of internal model control structure,the parameters are iteratively optimized by the objective function of the particle swarm optimization algorithm.Results of the theoretical analysis and simulation, which shows the output status of parameters u(t)and e(t)in the Internal Model Controller through the optimized parameters kp,ki and kd are more stable than before.(2)To improve the accuracy of the Fuzzy Smith Prediction Control System on the network delay prediction,to avoid delay estimation error occurred which can cause network system to appear "empty sampling" or "multi sampling" phenomenon,this paper proposes a prediction of the pure delay term tp on the closed loop link.When the network jitter is larger,tp has a great influence on the link delay,because the term tp can’t be eliminated on the denominator by the transfer function of the closed loop link the fuzzy Smith predictor control system.So increased the prediction model tpm for tp, which combined closed-loop link formula of Smith predictive control system to derivate process and simulation results,shows that the improved scheme can decrease the occurrence of the error by the delay of the link in a certain extent and improve the accuracy of estimates.(3)In order to reduce the packet loss rate of the RED algorithm,the improved fuzzy Smith predictor controller is proposed to improve the length and packet loss rate in the;RED algorithm.Because the packet loss rate of RED algorithm is affected by the length of the queue,so if we want to reduce the packet loss rate,we must control the length of the queue to maintain a relatively stable range.This paper makes full use of the better compensation control characteristics of the improved Fuzzy Smith Predictive Control System,to deal with the queue length of the RED algorithm by the controller.However, in the process of processing the queue length,the fuzzy rules in the control system can also calculate the lost packets,which lead to increased packet loss rate.To reduce the error,this paper introduces the link data packet’s survival time TTL as the particle weight of the PSO algorithm in the controller,so we have get a new formula for calculating the packet loss rate of RED.Theoretical analysis and experimental results show that in some extent the improved RED algorithm can reduce the network jitter, network delay,and the packet loss rate and improve the performance of the network. Finally,summarizing the main work of this paper,and put forward the next research plan,suggestions and views on the RED algorithm based on the Control Theory.