Optimal decentralized control systems with unreliable communication links

In traditional control systems such as industrial plants, automobiles, HVAC systems, etc. consist of a single dynamical system that is controlled by a single controller. Such systems are referred to as centralized control systems. Many of the modern control systems such as networked control systems, Internet of Things, mobile robotics, vehicular platoons, and smart grids consists of multiple dynamical systems that are controlled by multiple controllers. Such systems are referred to as decentralized control systems. The optimal design of centralized control systems is well understood. For centralized control systems with linear dynamics, quadratic cost, and Gaussian disturbances (abbreviated as LQG systems), the optimal control strategy is easy to synthesize and easy to implement. However, the optimal design of decentralized control systems is still an open area of research. In other words, there is no approach to obtain the optimal design of a general decentralized control system. Most of the result in the literature investigate models with specific features and exploit the features of the model to establish that the optimal control strategy is linear and has a finite dimensional representation. To be cont’d.

Faculty Supervisor:

Aditya Mahajan

Student:

Partner:

University of Southern California

Discipline:

Engineering

Sector:

Education

University:

McGill University

Program: