Research On Active Queue Management

Network congestion has become an important issue with the rapid development of the Internet. Further, AQM(Active Queue Management) is a hotspot in the current studies on TCP end-to-end congestion control. By evaluating the state of network and foretelling the congestion, it can drop the packet purposefully so that the sender side can be informed of the state of network and then adjust its sending rate. But the current algorithms are still not perfect in terms of response time, stability and sensitivity to the environment. This paper presents a PID (Proportional Integral Differential) AQM algorithm and validates it on NS2 (Network Simulator V2). Detailed research work includes the following:The research on the TCP flow model: Network congestion control is an interdisciplinary field between automatic control and computer science. With further research on the network congestion control, many flow control theories and network models have appeared. Based on fluid flow theory, Misra developed a dynamic model of the TCP congestion window with AQM in 2000. It is a relatively precise description of the TCP flow and has been accepted by many researchers. However, some flaws are inevitable because of its approximation in the model evolvement. This paper gives a detailed analysis of its shortcoming from both the theory and experimental results, and presents an improved model.Based on the improved model, this paper applies a new PID and PID-like controller design method to AQM controller. By designing an ideal D-stable region in the complex plane and setting the roots of the closed loop system within the region, a sound performance of the closed loop system is guaranteed.To carry out the experimental simulation, the function of NS2 is extended to support PID algorithm and meanwhile the general principle of the function extension of NS2 is shown. The experimental simulation results show that the new PID algorithm has a better overall performance.The research in this paper provides a valuable reference for the further study of network congestion control.