Research on Congestion Control Model and Algorithm for High-Speed Network Based on Genetic Neural Network and Intelligent PID

Abstract

In this paper, we have studied high-speed network congestion control problem from two levels of end system and communication subnet. First of all, a high-speed network congestion control model is presented based on HSTCP algorithm. From the aspect of control theory, we research AQM algorithm in communication subnet of high-speed network and give a design method of AQM controller based on genetic neural network and intelligent PID. In the system, we add two key parts into traditional PID controller. The first part is application of neural network, which is responsible for adjusting PID controller parameters online. The second part takes use of global convergence in genetic algorithm and sets up genetic neural network model to optimize weight coefficients and for neural network. In this paper, we integrate the advantages of genetic algorithm, neural network and PID control model, by which a high-speed network congestion control model is set up based on genetic neural network and PID, by which, a new AQM algorithm is designed for high-speed network based on HSTCP/AQM model that is called IPAQM. In this way, we have opened a new approach for foundation of high-speed network congestion control model and research of AQM algorithm.