Research Of Network Adaptive Congestion Control Algorithm In Active Queue Management System

In the development of high speed communication network, operational flow shown sudden and diversity to enhance the network quality of service to create more difficulties, arising from network congestion has became a bottleneck question restricting the development of the network. The continuous development of Active Queue Management (AQM) can not only solve the congestion problem, but as a supplementary TCP congestion control, QoS and the Internet can enhance the robustness of the research in the field of great significance. However, existing AQM algorithm in response to the speed and sensitivity of such areas as the environment is still deficient. Fuzzy control can be used to control the complexity, time-varying, uncertainty of the network, the fuzzy control can be introduced into the AQM congestion control of intelligent, AQM to improve the stability and robustness.This thesis focuses on the adaptive mechanisms to improve AQM based on fuzzy control algorithm, the main work is as follows:①Analyzing AQM research and significance, discussing of the Fuzzy Control in AQM in the direction of applied research and its improvement. Analyzing the existed AQM algorithm--REM and Fuzzy Control-based improvement algorithm of REM, discussing advantages and disadvantages of each.②Adaptive mechanism for FREM algorithm—ATFREM, the main idea is to use the average queue length as congestion, under the instruction of the actual network traffic load adaptive algorithm TQL, so network systems in a dynamic environment of the high-throughput, low-delay, low-dropping probability reaches a certain level of balance.③Adaptive mechanism for FREM algorithm—AFREM, the main idea is to use the average queue length and packet loss rate as a percentage of change in congestion Indication, FREM output of the probability of the discard dynamically adjusted, in a dynamic environment to control the queue length will be fixed near the TQLSimulation results have shown that these two control methods improves the performance of the congestion control system greatly, and makes a better trade-off between the adaptive performance and real-time performance.