Behrooz Rezaie

  PhD Thesis Defense Session

  Abstract

  In this thesis, we carry out global stability analysis, bifurcation study, and control design for a nonlinear delayed model of congestion control system in computer communication networks namely Kelly model. The model covers the system dynamics for a network including arbitrary number of sources and links. First, sufficient conditions for global stability of the system in steady state are derived by considering time-variability of some parameters such as link capacities and time-delays in Kelly model. Then, Hopf bifurcation analysis is studied for Kelly model of general networks with a delay parameter. By choosing the gain vector and the delay parameters as the bifurcation parameters, the conditions and properties of Hopf bifurcation are obtained. In the next part, Hopf bifurcation control is studied to postpone the onset of the bifurcation and thereby to avoid nonlinear dynamical behaviors. First, we propose a dynamic delayed feedback control scheme which consists of a combination of the delayed control feedback method and a washout filter. Then, an adaptive control method is proposed for tuning the control parameters such that the closed loop system copes with the effect of changes in the system parameters by updating the control gains. When the parameters of the system are not known, the problem can be solved by adding a parameter identification algorithm to the control scheme. Finally, the proposed control methods are numerically studied using simulation results for different scenarios. The simulation results show that the proposed control methods can effectively enhance the system behavior.

  Date: Momnday 22 Feb 2010

  Place: Classroom No. 101, Electrical Engineering department