Active Queue Management (AQM) aims to achieve high link utilization, low queuing delay and low loss rate in routers. However, it is difficult to adapt AQM parameters to constantly provide desirable transient and steady-state performance under highly dynamic network scenarios. They need to be a trade-off made between queuing delay and utilization. The queue size would become unstable when round-trip time or link capacity increases, or would be unnecessarily large when round-trip time or link capacity decreases. Effective ways of adapting AQM parameters to obtain good performance have remained a critical unsolved problem during the last fifteen years. This thesis firstly investigates existing AQM algorithms and their performance. Based on a previously developed dynamic model of TCP behaviour and a linear feedback model of TCP/RED, Auto-Parameterization RED (AP-RED) is proposed which unveils the mechanism of adapting RED parameters according to measurable network conditions. Another algorithm of Statistical Tuning RED (ST-RED) is developed for systematically tuning four key RED parameters to control the local stability in response to the detected change in the variance of the queue size. Under variable network scenarios like round-trip time, link capacity and traffic load, no manual parameter configuration is needed. The proposed ST-RED can adjust corresponding parameters rapidly to maintain stable performance and keep queuing delay as low as possible. Thus the sensitivity of RED's performance to different network scenarios is removed. This Statistical Tuning algorithm can be applied to a PI controller for AQM and a Statistical Tuning PI (ST-PI) controller is also developed. The implementation of ST-RED and ST-PI is relatively straightforward. Simulation results demonstrate the feasibility of ST-RED and ST-PI and their capabilities to provide desirable transient and steady-state performance under extensively varying network conditions.

The optimal steady-state control, and suboptimal adaptive control, of disturbed single-input-output systems are introduced, and the class of systems considered is defined. It is noted that the stochastic tracking problem divides into a deterministic tracking problem and a stochastic regulator problem; the solutions to these two problems are shown to be independent but formally similar. The continuous regulator problem is approached via both frequency and time domain methods: the former method is extended to cover unstable systems; the latter method is extended to include systems with input delay. The two regulators are shown to be externally equivalent. The frequency domain method is briefly described for discrete systems, and shown to include the minimum variance regulator of Åström and Peterka as a special case. Some systems which allow measurement noise to be treated as a system disturbance for the purposes of optimal controller design are investigated. A novel class of control laws is described in both continuous and discrete time; in the same way as the minimum variance regulator forms the basis of the self-tuning regulator of Åström and Wittenmark, these minimum variance controllers from the basis of a self-tuning controller. These minimum variance controllers have a number of advantages over the minimum-variance regulator, and are open to a number of interpretations including: a model following control law, and an extension of classical control laws to systems with delay. The optimality of this class of control laws is investigated, and analogies drawn with the previously considered k-step-ahead control laws; some examples are given to illustrate the method. An adaptive control law combining the above minimum variance controllers with a linear least-squares algorithm is proposed and shown to be self-tuning. These self-tuning controllers are only slightly more complex than the self-tuning regulator of Åström and Wittenmark, but have a number of advantages. Intuitive justification is given for the conjecture that some methods of Ljung, developed for the analysis of the self-tuning regulator, are applicable to the self-tuning controller. Simulated examples are given which compare and contrast the performance of the self-tuning controller with that of the self-tuning regulator. The first steps towards a quasi-continuous self-tuning controller are outlined.

This paper deals with the consensus problem under communication network inducing delays. It is well-known that introducing a delay leads in general to a reduction of the performance or to instability due to the fact that timedelay systems are infinite dimensional. For instance, the set of initial conditions of a time-delay system is not a vector but a function taken in an interval. Therefore, investigating the effect of time-delays in the consensus problem is an important issue. In the present paper, we assume that each agent receives instantaneously its own state information but receives the state information from its neighbors after a constant delay. Two stability criteria are provided based on the frequency approach and on Lyapunov-Krasovskii techniques given in terms of LMI. An analytic expression of the consensus equilibrium which depends on the delay and on the initial conditions taken in an interval is derived. The efficiency of the method is tested for different network communication schemes. / <p>QC 20110120</p>