A Distributed Algorithm with Event-Triggered Communication for Resource Allocation Problem

Abstract

This paper focuses on the distributed resource allocation problem in continuous-time multi-agent systems over undirected networks, in which each agent has a local convex cost function only known by itself and all agents are required to converge to the global optimizer of the sum of the local cost functions under constraints. A gradient-based distributed algorithm is proposed for solving the problem. Furthermore, the step-size of gradient is a constant. A sufficient condition is given which guarantees that the proposed algorithm can achieve the global optimization. Moreover, in order to avoid continuous communication, an event-triggered communication control strategy is designed. Finally, an example of the economic dispatch problem in smart grid is presented to validate and illustrate the feasibility of the proposed algorithm.