Robust Dynamic Average Consensus for a Network of Agents with Time-varying Reference Signals

TL;DR

This paper introduces robust, smooth, edge-based dynamic average consensus algorithms for networks of agents with time-varying signals, ensuring zero steady-state error and robustness to network changes.

Contribution

It proposes novel consensus algorithms that are robust, avoid chattering, and guarantee convergence in dynamic, changing networks with local computations.

Findings

Algorithms achieve asymptotic convergence to the average of reference signals.

Simulation results validate robustness and performance improvements.

Designed for balanced, strongly connected communication graphs.

Abstract

This paper presents continuous dynamic average consensus (DAC) algorithms for a group of agents to estimate the average of their time-varying reference signals cooperatively. We propose consensus algorithms that are robust to agents joining and leaving the network, at the same time, avoid the chattering phenomena and guarantee zero steady-state consensus error. Our algorithms are edge-based protocols with smooth functions in their internal structure to avoid the chattering effect. Furthermore, each agent is only capable of performing local computations and can only communicate with its local neighbors. For a balanced and strongly connected underlying communication graph, we provide the convergence analysis to determine the consensus design parameters that guarantee the agents' estimate of their average to asymptotically converge to the average of the time-varying reference signals of the agents. We provide simulation results to validate the proposed consensus algorithms and to perform a performance comparison of the proposed algorithms to existing algorithms in the literature.