An application of Genetic Fuzzy System in Active Queue Management for TCP/IP multiple congestion networks

Phuong Huy Nguyen, Thi Mai Thuong Duong, Thu May Duong

In this paper, authors proposed a new algorithm to improve the Random Exponential Marking (REM) on TCP/IP network by combining Fuzzy System and Modified Genetic Algorithm (MGA). This new approach allows automatic adjustment of fuzzy parameters according to the network dynamics. Using NS-2 simulator, the results are verified and compared to some traditional AQM algorithms on a multiple-congested TCP/IP network.