Control methods for data flow in communication networks.

摘要

In this dissertation, we investigate various control methods for data flow in communication networks. First, we develop a rate-based flow controller for self-similar network traffic, which consists of a robust Hinfinity control block and an adaptive LMMSE capacity predictor. The controller guarantees robust stability against time-varying time delay uncertainties and improves the transient response by predicting the self-similar cross traffic. Window-based congestion control methods are also explored for TCP traffic on IP networks. We propose a variable structure approach in Active Queue Management (AQM) support Explicit Congestion Notification (ECN). By analyzing the robustness and performance of the control scheme for the nonlinear TCP/AQM model, we show that the proposed design has good performance and robustness with respect to the uncertainties of the round-trip time (RTT) and the number of active TCP sessions, which are central to the notion of AQM. Alternatively, we design robust Hinfinity AQM controllers for the linearized TCP/AQM model, with the presence of uncertain time delays. The Hinfinity performance is analyzed and a switching control scheme is introduced to improve the system performance. Motivated by the Linear Parameter Varying (LPV) nature of the linearized TCP/AQM model, a switching Hinfinity control method is further investigated for LPV systems where we provide some stability conditions in terms of the dwell time and the average dwell time.