An Adaptive Target Queue Length FREM Algorithm

Fuzzy Control has been a research highlight in network congestion control field. But many existed fuzzy algorithm´s parameters and fuzzy rules are static, so it have defect in respond speed, stability and environment sensitivity. As to these problems mentioned above, based on the existed FREM (Fuzzy Random Exponential Marking) algorithm, this thesis explores adaptive FREM algorithm - AFREM (adaptive fuzzy REM) algorithm, the main idea is adjust the dropping probability of FREM algorithm according to the network load. We show theory analysis and simulation test. The test shows that AFREM adapts control the packets dropping according to flow. The new algorithm shows superior performance than FREM in controlling the queue length of the router and can achieve perfect control properties in the event of network dynamic change.