Research On Active Queue Management Algorithm Based On State Space

The fluid-flow model plays an important role in the area of network congestion control by describing the TCP/AQM system as a time-varying, nonlinear system with large time lag. However, this model can not approximate the system to adequate accuracy when the packet dropping probability P<<1while running with the packet-discarding scheme. It is thus of great significance to improve the fluid-flow model to deal with the shortage mentioned above.In this paper, the fluid-flow model is modified by analyzing the dynamic characteristics of the TCP/AQM system with packet-discarding scheme. And new AQM algorithms are proposed based on the modified model, by using state feedback control theory, output feedback control theory and Linear Matrix Inequality Method(LMI). Major results are listed as follows:(1) A new dynamic nonlinear model FFI for TCP/AQM system is derived by modifying the fluid-flow model. Accoording to the analysis of the TCP/AQM system’s dynamic charactermistics with the fluid flow theory. This new model is focused on solving the mismatch between an actual system running the packet-discarding scheme and the traditional fluid-flow model. Simulation results show that the FFI model approximates closer to the actual congestion control system running with packet-discarding scheme.(2) A novel AQM scheme named state feedback control algorithm ISFC based on FFI model is proposed. The FFI model is linearized using small signal analysis, then the state space model is established and a novel active queue management algorithm is designed using state feedback control based on this model. According to the theory of Lyapunov, the conditions for asymptotical stability of the system and the parameters of controller are given by using LMI.(3) An AQM scheme named output feedback control algorithm IOFC based on FFI model is proposed. The algorithm is designed using state feedback control based on the state space model of FFI with conditions for asymptotical stability of the system. And the controller design for IOFC is also given by using LMI.NS2simulation results show that, compared with traditional AQM algorithms, the ISFC and IOFC are more stable and robust.