Active Queue Management Algorithm For TCP Networks Based On Minimax Theory

The development of the Internet has brought convenience to people's lives.Meanwhile,with the increasing of the internet users and applications,the network congestion problem becomes more and more serious.Congestion is caused by the accumulation of data at the bottleneck of the router,such that the sender data cannot be sent to the receiver,which will result in the data loss and degraded network service quality.The router active queue management algorithm combined with the TCP control protocol is a wide and effective method to solve the current TCP network congestion control problem.Active queue management can effectively improve link utilization and reduce queue latency,which helps to enhance network service quality.Therefore,the research on the active queue management algorithm has become a hot topic.Aiming at the congestion problem existing in TCP network,the active queue management algorithm is deeply studied from the perspective of control theory in this dissertation.The time-varying delay,system parameter variation,nonlinearity and non-responsive UDP flow interference become the main point in this dissertation.The main research contents of this dissertation are given as follows:For the dynamic model of linearized TCP network,an active queue management algorithm based on Minimax theory is investigated.In the design process,the linear matrix inequality method is used,the time-varying delay existed in the states is fully considered,and the link capacity error is regarded as the interference.A test function is established by combining the Lyapounov-Krasovskii functional.The maximum interference of the system is obtained by using the Minimax method.The artificial scaling and estimation of the interference items are avoided,which effectively reduces the conservativeness.Meanwhile,the state feedback controller and the output feedback controller are designed.Moreover,the simulation results illustrate the effectiveness of the proposed method.The active queue management problem of network congestion is addressed by combining Backstepping method and Minimax theory for the dynamic model of TCP nonlinear network system.In this model,the rate and queue length are regarded as the state variables,respectively.During the design process,the system is not linearized.Furthermore,the Minimax method is combined with the backstepping to deal with the disturbances,which can be eliminate the sensitivity on the disturbance.With the help of comparing the different method,the effectiveness of the presented scheme can be verified.Under the condition of existing the maximum interference,the design problem of the robust controller is addressed for the TCP/AQM network congestion system.In order to eliminate the steady-state error,the integral term function is introduced,and the robust controller is obtained to make the nonlinear network asymptotically stable and the packet loss small.Next,the sufficient condition on the choices of the controller parameters is given.The superiority of the design method is illustrated by simulation analysis.Consider the robust adaptive controller design for the TCP/AQM nonlinear network system.The window and queue are regarded as the state variables.In the case of parameter uncertainty and UDP flow interference,the Minimax theory and adaptive backstepping method are used to design the AQM algorithm.During the process of designing adaptive laws,the parameter mapping mechanism is added.Via the constraint of the piecewise function mapping of the estimated value and the auxiliary variable,the parameter estimation process always occurs within the specified range,which can ensure that the uncertainty parameter is changed within the specified range and the efficiency of error tracking is improved.The active queue management controller is designed based on the adaptive disturbance rejection algorithm.The parameter uncertainty is obtained through the mapping relationship between TCP network system window and rate.A modified method is proposed to avoid the discontinuity of the piecewise function.And a new adaptive congestion controller is designed.With the aid of introducing the adaptive parameter reconstruction mechanism,the smoothness of the designed controller is guaranteed.Based on the comparison results,the superiority of the designed Minimax adaptive interference suppression algorithm is illustrated.Lastly,the summary of the whole dissertation is given and the research directions in future are put forward.