Researches On Active Queue Management With Network Simulator Version 2

Nowadays, going with the development of the Internet and sophisticated multimedia technology, the increasing user's number and their requirement bring serious load problems. When the network exists excessively packets, and network resources (for example:gateway handling capacity, buffer size, link bandwidth) is limited, it will cause the network congestion which led to a sharp decline in the throughput of the network and loss massive transporting packets. Therefore congestion algorithms needed to adjust the message transmission rate, ease congestion and then to improve quality of service of the network.According to occurring position, congestion control algorithm is divided into two categories:the link algorithm and the source algorithm. In the source algorithm, Transmission Control Protocol is the most widly used as an indispensable factor of guaranteeing network robustness. In the link algorithm, Active Queue Management (AQM) is focus of current research. The Random Early Detection (i.e. RED) algorithm is a representative AQM algorithm, and is also the only candidate algorithm recommended by RFC2309. It is widely attented and researched by scholar.Firstly, we give introduction the researching status of the congestion control algorithms, and the problemes of the congestion control algorithms at present. We expatiate the definition of congestion, analyse the causations of congestion. We introduse what is congestion control and the experimental tool of this paper, and elaborate the queue mechanism of router in detail. We also introduce the Network Simulator version 2 the experimental tool of this paper.In the second section we summarize the RED algorithm and its revised algorithms, such as ARED, SRED, FRED and BLUE. According to the modelling thought of the citation, we deduce the relational expression between average queue size and packet loss rate. Analyzed the relational expression, the nonlinear character between average queue size and packet loss rate is much more matched. That provides an theoretical basis for improved RED algorithms in this paper.In this paper, is related we make the improvement of RED algorithm for nonlinear studies, proposed two nonlinear RED algorithm, the last use of network simulation tools to simulate and verify the validity of the algorithm.Using the nonlinear character between average queue size and packet loss rate, we present two improved RED algorithm named MRED and JRED. The core expression of the two improved algorithms are also given. NS2 is our experimental. Simulation studies show that the improved algorithms increase the average throughput, decrease packet loss rate and enhances stableness and reliability of network. Compared the MRED and the JRED, the JRED algorithm's overall performance is much more preferable.Finally, we give a summarization of this paper and present the work we will do in the future.