On Active Queue Management Algorithms For Network Congestion Control

With the development of Internet based on TCP/IP, the scale, users and traffics have experienced an explosive growth since 1990's. The network congestion has become more serious and complex due to the ever-increasing network application types and dynamic network parameters such as active sessions and round trip time. Congestion often results in decline of quality of service (QoS) in terms of transmission delay and throughput, while the network resource utilization like bandwidth and buffers are also affected seriously. The congestion control is always a hot spot in the field of network research.Internet primarily relies on TCP end-to-end congestion control to avoid network congestion. However TCP strategies could not meet many various demands of every application on the complex network. In fact, it will be more effective for detecting and preventing congestion if the routers perform queue management schemes. Congestion control strategies for the routers have been discussed widely. There is also a new method of congestion control which acts on application layer. From the high application level, using sliding mode control and artificial neural network theory for flow control can avoid or lighten the congestion in computer network. It is a new field of the research on congestion control.Congestion control strategies for IP based networks are introduced in detail from two different layers of routers and end host in this thesis. Systematic research to meet practical requirements of IP based networks is performed in depth. The main research works and conclusions are as follows:The concept and meaning of congestion and congestion control is introduced. TCP and IP congestion control mechanism is analyzed particularly. The ordinary algorithm of TCP and IP congestion control, as well as the existent problems and improved measures of them are analyzed and summarized.In this paper, we integrate the thinking of controlling theory into the theory of neural network, and bring forward new AQM scheme based on traditional AQM algorithm. Simulation results indicate that the new algorithm is effective. We designed a single neuron self adaptive PID controller. And the simulation results indicate the capability such as adjust time, error bound and stability of queue size in Router is superior to the traditional PID controller.In sliding mode control system, the sliding mode possesses strong robustness, which is very suitable for complex time-varying TCP network system. One AQM algorithms are designed based on sliding mode control for the continual network systems and discrete-time network systems with matched uncertainties. An asymptotically stable sliding surface is designed by pole placement. The sufficient condition is obtained for the existence of the sliding surface. The design of sliding surface compensates for the effect of uncertainties. Simulation results demonstrate that the proposed controllers have good stability and robustness with respect to the uncertainties of the number of active TCP sessions, link capacity and the round-trip time (RTT).The conclusion is drawn for the whole dissertation, and the further research aspect is put forward.