Research And Improvement Of Red Algorithm Based On Active Queue Management

With the continuous development of the Internet and the growing emergence of new network applications, especially the increases of multimedia flow such as the voice and the video, the network information flow has been increasing explosively which causes the fact that the broadband resources become nervous and the network congestion becomes more and more seriously. On the other hand, the demand of the quality of the network service (Qos) becomes higher and higher, such as the transmission delay and the throughput. In this case, the preventing and controlling of network congestion has been a hot issue in the field of network research at home and abroad in recent years.There are two kinds of congestion control mechanism. One is the TCP protocol control based on the source port and another is the congestion control based on intermediate node (router). Routers can not only effectively monitor the length of real-time queue, but also examine the influence of various flows on congestion, so as to notify the source port of the flow to adjust the network. It is thus clear that the congestion control mechanism based on intermediate node (router) is a very effective solution.Active queue management (AQM) is the most prominent in congestion control mechanisms based on intermediate node. In this paper, we systematically evaluate several typical AQM algorithms and focus on the only AQM candidate strategy RED (Random Early Detection) recommended by IETF. RED algorithm has resolved the problems brought by network traffic burst. However, it remains some shortcomings in the aspect of algorithm stability and parameter sensitivity. So, it is extensively studied at home and abroad. In this paper, a nonlinear adaptive algorithm (NLARED) is proposed on the basis of RED algorithm. There are two major improvements:first, on the calculation of packet loss rate, the partial large Cauchy distribution of fuzzy membership function is used to replace the original linear piecewise function. Second, the adaptive adjustment Pmax mechanism is added into the algorithm. The average queue length is applied to reflect the utilization of buffer resources, so as to determine whether the congestion control index is moderate or not. For different section, different reference quantity is introduced to adjust Pmax, so as to make the adjustment of Pmax more accurate and timely and avoid introducing new static parameters at the same time.The NS2simulation experiments show that NLARED algorithm can effectively adapt to the change of network traffic, maintain the stability of the length of queue and reduce the parameter sensitivity. It has obvious improvement in keeping low packet loss rate and high throughput and has better performance than the RED algorithm.