Based On Active Queue Management Congestion Control Strategy

In this day and age, with the rapid development of the Internet, more and more people tend to retrieve diverse information via Internet. However, as users and applications growing, the subsequent dramatic increase in traffic will result in much more heavier congestion over the Internet, leading to an obvious decline to the network performance. Therefore, the network congestion control has become an important step to enhance the performance of the Internet.Different from the TCP-based congestion schemes, Active Queue Management (AQM) strategy gives full play to the network's ability in smoothing congestion, which achieved good results. This thesis analyzes the internal structure of the Adaptive RED algorithm by the way of time-domain analysis and redesigns a new discrete controller according to the result of analysis, and then applies the new controller to the AQM algorithm, which conducts a more stable queue for routers. The main contributions of this paper are listed as follows:(1) It firstly made a detailed introduction to the classic RED & ARED algorithm, and calculated new parameters for PI controller according to the requirements of the follow-up exper(?)ments.(2) The 1-order difference equation, which used to depict the system of average queue length in ARED algorithm, was converted to the relative pulse transfer function and the relative continuous transfer function respectively. Time-domain analysis was used to analyze this transfer function. Thus, a new approach to improve the ARED algorithm was proposed, that is, using a 2-order difference equation to replace the 1-order difference equation.(3)On the basis of the above analysis, this paper determined the design goals for the 2-order system, and then got the 2-order transfer function. Thus, we used the time-domain analysis to examine the dynamic nature of the transfer function, and then converted it to the corresponding 2-order difference equation. Finally, the simulation results verify the performance of the improved algorithm.