On a Potential Game-theoretic Approach to Event-triggered Distributed Resource Allocation

TL;DR

This paper introduces a potential game-theoretic framework for event-triggered distributed resource allocation in multi-agent systems, ensuring efficient coordination and avoiding Zeno-behavior through derived conditions.

Contribution

It presents a novel potential game approach with a linear parameter-varying dynamic, linking consensus protocols to resource allocation and establishing minimum inter-sampling time bounds.

Findings

Dynamic resembles a consensus protocol

Equivalent to the replicator dynamic

Conditions for minimum inter-sampling time

Abstract

This paper proposes a potential game theoretic approach to address event-triggered distributed resource allocation in multi-agent systems. The fitness dynamic of the population is proposed and exploited as a linear parametervarying dynamic to achieve the event-triggered distributed resource allocation. Firstly, it is shown that the dynamic resembles a consensus protocol and is equivalent to the replicator dynamic. Further, conditions ensuring a minimum time bound on the inter-sampling intervals to avoid Zeno-behavior have been derived. An economic dispatch problem is presented to illustrate the theoretical results.