AQM Controller Research Based On Optimal Control Method

If the traditional "Drop-tail" algorithm is used on the router, when the queue length exceeds the buffer of the queue, the arrival packets are discarded and only when the queue buffer can accommodate the new packets are allowed to enter the queue. This algorithm has many shortcomings: it leads to queue length jitter amplitude, leading to unpredictable long packet queuing delay and link utilization is low.One way to solve the defects of the algorithm is to use the AQM (Active Queue Management) algorithm. The AQM (Active Queue Management) algorithm recommended by the IETF is the router end congestion control technique, the most important way to solve the issues related to control congestion on the Internet is the combination of TCP congestion control and AQM.AQM algorithm has long been a hot research area for the Internet router end congestion control. As much as possible between the data flow rationalization, fair allocation of network resources, packet delay, the highest throughput on link, and packet discard to minimize, etc. These are the ultimate goals of the implementation of congestion control mechanism.Established on the basis of TCP and AQM congestion control, combined with the knowledge of control theory, we use the optimal control method proposed a new AQM controller. Through theoretical analysis and simulation proved that this controller is better than the original part of the AQM algorithm in some respects.The main work of this paper: KELLY TCP congestion control model conversion (variant), a new system model, this model through the equation of state, the linear quadratic equation of state. Use the linear quadratic optimal control theory, the principle of minimal value, a necessary condition for the optimal solution to find the optimal control system performance indicators to a minimum. The optimal control parameters, k1and k calculated packet loss rate (a new AQM controller)-AQM controller based on optimal control methods.