Yugeng Xi

Distributed model predictive control for large-scale systems

This paper presents a novel distributed model predictive control algorithm based on Nash optimality, in which the online optimization of the whole system is decomposed into that of several small cooperative agents in a distributed way, thus can greatly reduce the computational complexity of large-scale systems. The convergence analysis of this distributed algorithm is also given.

Linear generalized synchronization of continuous-time chaotic systems

Abstract This paper develops a general approach for constructing a response system to implement linear generalized synchronization (GS) with the drive continuous-time chaotic system. Some sufficient conditions of global asymptotic linear GS between the drive and response continuous-time chaotic systems are attained from rigorously modern control theory. Finally, we take Chua’s circuit as an example for illustration and verification.

GLOBAL SYNCHRONIZATION OF A CLASS OF CHAOTIC SYSTEMS WITH A SCALAR TRANSMITTED SIGNAL

This letter proposes a new global synchronization theorem for a class of chaotic systems. Specially, in the synchronization theorem neither the linear part nor the nonlinear part of the chaotic system requires the special structure indicated in [Wang & Wang, 1998]. We take a linear combination of the original system state variables as the scale-driving signal. We prove that the global synchronization can be attained through the simple linear output error feedback. The linear output error feedback gain is a function of a free parameter. We use Chuas chaotic oscillator and Saitos hyperchaotic system to illustrate the applicability of our synchronization scheme and discuss the issues concerning robustness.

Bifurcation analysis of an Internet congestion control model

We investigate the bifurcation behavior of a discrete-time model of Internet congestion control system, which stands for a single link shared by multiple sources. By choosing the gain parameter and the number of sources as bifurcation parameters, we prove the existence of period doubling bifurcation in the model and study the stability of bifurcation. We also find that chaotic behavior of the model can be controlled by limiting the amplitudes of the sending rates.

Internet congestion control with delayed state feedback

Random early detection (RED) has been proposed to avoid impending Internet congestion and simultaneously achieve high throughput and low average delay. However, it has been found that the interaction between a RED gateway and TCP connections may exhibit complicated nonlinear (in particular, chaotic) behavior and lead to network instability. In this paper a time-delayed state feedback controller is implemented to enhance the stability of RED. We validate the algorithm by extensive ns-2 simulations.