On Problems Of Congestion Control Of TCP/IP Networks

Nowdays, with the rapid development of science and technology, the appearance of varied network applications made network congestion control a significant problem. The network congestion became a bottleneck problem, which restricts the development and application of networks. Congestion often results in the decline of QoS (Quality of Service) parameters such as transmission delay and throughput. The network per. formance and resource utilization are also affected seriously, and effective QoS guarantee cannot be provided accordingly. Therefore, it is very important to solve the congestion problem effectively for improving the network performance. The congestion control problem has become an active issue in the field of computer network research.Based on the analysis of congestion control mechanism of TCP/IP, the congestion control problem is dealt with by means of the methods in control theory and economy field. The reseach work is summarized as follows.Considering the time varying and uncertain characters of networks, an active queue management (AQM) algorithm, i.e. IMC.Smith algorithm is proposed based on the combining of inner model controller with Smith estimate controller. The simulation results of the algorithm under different instances show the superiority of the IMC-Smith algorithm.Considering the time delay of networks and the weakness of depending on precise model of the existing algorithms, an active queue management algorithm aimed at large delay dynamic network is proposed based on the combining of fuzzy controller with improved Smith estimate controller. The effect of large time delay to queue stability may be compensated by Smith estimate controller, the TCP congestion can be controlled timely. And the defect of depending on precision model of Smith estimate controller can be overcame by fuzzy controller which may work well without the precision model of control object, and the fuzzy controller has good robustness to dynamic network and suitable for network with incertitude time delay.A virtual rate algorithm (VRC) is proposed based on the combining of gray prediction, adaptive control and virtual rate algorithm. The disadvantages of slow responding speed of the algorithms based on queue are resolved by the improved virtual rate algorithm. And the control parameters can realize adaptive adjusting online to ensure the system stability on balance point to overcome the difficulty of adaptive adjusting parameters of traditional PID controllers. The system may obtain better transient response and steady characteristic by this algorithm.A new congestion control method with explicit rate feedback is proposed based on the combining of traditional Smith estimate controller and filter. With the new method, the effect of transmit delay of TCP network is overcame, the TCP sending window may quickly response the change of the network load to avoid the grouping losing, the flux of the network is more steady. The theoretic analyses indicate that this method may ensure system robustness with model matching or un-matching.The pricing policy in economy is applied to the problem of network congestion control. Firstly the resource possess ratio is introduced into pricing control to construct the pricing strategy. Then the pricing control algorithm based on Stackelberg strategy is presented, which combines the utility function model of network with Stackelberg strategy in the game theory. The simulation results show the superiority and the effectiveness of the algorithm.Considering the control algorithm advanced based on optimize theory and it's commonly analysis frame, the stability of the algorithm under a simplify network condition is proposed. The global stability of the congestion control algorithm under the commonly topology is proved without regarding of the network transfer delay; the condition of the global stability of the congestion control algorithm with the single link topology is proposed considering the network transfer delay. These conclusions provide theory basic for network design, ensuring network stability and avoiding network congestion.Lastly, the reseach work is concluded. The further research direction is pointed out, and the prospect of solving the congestion control is looked forward.