Model-based Network Congestion Control

With the rapid development of network users and all kinds of applications, Internet traffic is becoming extrodinary busier nowadays. Congestion phenomenan occurs easily due to the restriction of Internet resource bottlenecks and the bursty nature of Internet traffic. So how to prevent and control congestion is one of the most active fields in the computer networks.The backgrounds and causes of Internet congestion phenomenon are introduced in the beginning of this thesis. This thesis also describes and compares some significant TCP/IP congestion control algorisms in detail, and researches some novel research methods and results in the Internet congestion control with the perspective of classic control system. The study in this thesis is listed as follows:1. Because RED and its modified algorithm depend on the intuition, they are lack of systematical and theoretical, the thesis inferres the simplification model based on the fluid stream theory proposed by on Misra V has been.2. Based on this model, the P controller is applied in the congestion control. The thesis compares with the queue curve under the different K condition, and carries on the frequency range analysis.3. In view of the system existing static error under the P control, the PI controller is introduced. The thesis has compared with the P controller through the simulation curve.4. In view of the system static error reducing under the PI control, but the system speed of response reducing, the PD controller is introduced. Control performance caused by PI and PD controller has been carried on the comparison through the simulation curve.5. The experiment indicates the PD controller can reduce the overmodulation of the system, improve the speed of response, but the queue curve vibrates.6. According to the loss rate of data packet is dynamic, the PI controller needs adjust the parameter dynamically to guarantee the algorithm has the quicker convergence rate and the smaller vibration, and fast response to the network change, enhancing the use factor of the buffer. The auto-adapted AQM algorithm—API-V controller is purposed. The theoretical analysis and the simulation indicate that the performance of the API-V algorithm surpasses the PI algorithm and its modified algorithm under the dynamic network environment.