Distributed Optimization and Learning

Estimating tolerable communication delays for distributed optimization problems in control of heterogeneous multi-agent systems

This study addresses the problem of distributed optimization in control of heterogeneous linear multi-agent systems. Instead of assuming a perfect communication medium as in many existing approaches, we consider two important issues arising from communication networks. First, we assume the existence of communication delays when each agent receives information from its neighbors. Second, since communication networks are generally unreliable, we assume that each agent interacts with other agents through random digraphs. Finally, we derive delay-dependent sufficient conditions in the form of linear matrix inequalities (LMIs) to prove convergence to optimal solutions.

Fault estimation in networked systems

A distributed fault estimation approach for a class of continuous-time nonlinear networked systems subject to communication delays

In this study, the problem of the fault estimation controller design is investigated for a class of nonlinear networked systems over a communication topology. In real systems, due to unreliable communication channels, data may be delayed or lost when it is exchanged between an agent and its neighbors. To explore this concern, in the addressed approach, each agent utilizes an augmented system based on a given communication topology for estimating the fault and the states, both within itself and in its neighbors, where information from neighbors is received with a time delay. Delay-dependent conditions in the form of linear matrix inequities (LMIs) are derived to guarantee the stability and to obtain the parameter matrices of the observer. Eventually, the simulation results are included to illustrate the performance of the addressed fault estimation approach.