A Consensus Protocol under Directed Communications with Two Time Delays and Delay Scheduling

TL;DR

This paper analyzes the stability of a second-order consensus protocol with directed communication affected by two unknown, constant, and independent delays, introducing a novel delay scheduling method to enhance system performance.

Contribution

It introduces the Spectral Delay Space for stability boundary depiction and the Delay Scheduling concept to improve consensus speed by strategically increasing delays.

Findings

Exact stability boundaries are characterized in the delay domain.

Delay scheduling can paradoxically improve system performance.

Spectral Delay Space simplifies stability analysis with multiple delays.

Abstract

This paper studies a consensus protocol over a group of agents driven by second order dynamics. The communication among members of the group is assumed to be directed and affected by two rationally independent time delays, one in the position and the other in the velocity information channels. These delays are unknown but considered to be constant and uniform throughout the system. The stability of the consensus protocol is studied using a simplifying factorization procedure and deploying the Cluster Treatment of Characteristic Roots (CTCR) paradigm. This effort results in a unique depiction of the exact stability boundaries in the domain of the delays. CTCR requires the knowledge of the potential stability switching loci exhaustively within this domain. The creation of these loci is an important contribution of this work. It is done in a new surrogate coordinate system, called the \emph{Spectral Delay Space (SDS)}. The relative stability, i.e., the speed to reach consensus of the system is also investigated for this class of systems. Based on the outcome of this effort, a paradoxical control design concept is introduced. It is called the \emph{Delay Scheduling}, which is another key contribution of this paper. It reveals that the performance of the system may be improved by increasing the delays. The amount of increase, however, is only revealed by the CTCR. Example case studies are presented to verify the underlying analytical derivations.