Research On Network Model And Congestion Control Mechanism Based On Control Of Transmission Rate

Networks had little layers and simple structure when it was constructed at the beginning. And also there are almost no meaningful control problems exist in network systems. With the high-speed computer network technology development, network structure becomes complex day by day, information exchange has been strengthened continuously and the degree of automation has been improved quickly. Formerly separated various activities obtain the conformity under the information flow function. In this situation, the traditional heuristic study has been unable to meet the needs of network congestion control. Therefore, using the control theory to analyze and research network congestion becomes a hot topic rapidly. It makes us to know the network system and the network congestion phenomenon from more complete and comprehensive direction.The paper discusses the network congestion control based on the network which transmits information with best-effort service. Without fundamental changing in the current network, the research target is to enhance the network control performance through improving the existing congestion control algorithms and designing new control algorithms. The main research contents have: (1) Analysis of the current computer network congestion and congestion control objectives; (2) Illustrate the modeling principle and process of the network simulation, studies the simulation model based on real object, and gives the basal structure of three layers network simulation model; (3) Research for computer network simulation based on the OPNET12.0 network simulation software, model and analyze the CSMA/CD protocol; (4) A double-end cooperative congestion control mechanism is proposed and analyzed. And we have studied the performance of the control mechanism via simulations on OPNET software. Simulations show that the mechanism can improve network performance under random propagation delay. (5) Proposed improvement algorithm PID-RED, and compared it with RED algorithm. The simulations show PID-RED algorithm has good performance than RED does.The research is supported by open foundation of the Field Bus Technology & Automation the key lab of Beijing. It is one of the theory researches in the Network Automation team. I appreciate the careful guidance of Professor Sun Dehui, Dr. Shi Yuntao, and Dr. Li Zhijun.