Distributionally Robust Cascading Risk Quantification in Multi-Agent Rendezvous: Effects of Time Delay and Network Connectivity

TL;DR

This paper introduces a distributionally robust risk framework for multi-agent rendezvous, analyzing how network connectivity, delays, and noise influence cascading failures, with validated theoretical and simulation results.

Contribution

It presents a novel distributionally robust risk measure for cascading failures in multi-agent rendezvous considering delays and network effects, with closed-form expressions.

Findings

Risk increases with communication delays and noise.

Network topology significantly affects cascading failure risk.

The framework provides actionable insights for resilient network design.

Abstract

Achieving safety in autonomous multi-agent systems, particularly in time-critical tasks like rendezvous, is a critical challenge. In this paper, we propose a distributionally robust risk framework for analyzing cascading failures in multi-agent rendezvous. To capture the complex interactions between network connectivity, system dynamics, and communication delays, we use a time-delayed network model as a benchmark. We introduce a conditional distributionally robust functional to quantify cascading effects between agents, utilizing a bi-variate normal distribution. Our approach yields closed-form risk expressions that reveal the impact of time delay, noise statistics, communication topology, and failure modes on rendezvous risk. The insights derived inform the design of resilient networks that mitigate the risk of cascading failures. We validate our theoretical results through extensive simulations, demonstrating the effectiveness of our framework.