Robust Aqm Algorithm Based On Sliding Mode Variable Structure Control

With the increasing level of information, Internet technology has been closely interrelated to people's lives. On the one hand the number of Internet users experience explosive growth, on the other hand many new services such as voice, video, and other multimedia streams appeared for the passed few years. But the growth of data traffic causes a great burden to the routing node. As a result, the Internet quality of service has been badly influenced, which result in more serious network congestion.AQM algorithm is the best way to solve the network congestion problem, this paper proposes several AQM algorithms focused on the stability and robustness of network. The contributions of this thesis are as follows:(1) Because of many uncertain factors in the network, an AQM algorithm based on variable structure control is proposed. VSC algorithm which is simply designed, easy to realize, not depended on the object model has strong robustness to those uncertain factors, and be good at nonlinear control. The simulation results show that the algorithm has better stability, small network latency, and robustness to the changes of the network.(2) In order to weaken the jitter of the VSC controller, this paper proposes two new AQM algorithms based on intelligent VSC. FVSC and NVSC algorithm which take advantage of fuzzy reasoning and the gradient learning of the single neuron adjust the controller parameter. They inherit the merits of the VSC algorithm and also reduce the jitter. Simulation results show that these two algorithms not only retain the good performance of the VSC algorithm, but also keep better queue length in the same network environment.(3) Because the network cannot build precise mathematic model, an AQM algorithm based on adaptive single neuron and variable structure control is proposed for the situation. The corrected control is designed to make the state vector arrive the sliding surface and weaken the jitter, and the equivalent control which outputs from the single neuron is adaptive to the uncertainties. Simulation results prove that the ANVSC algorithm show good stability and robustness in different network environment.