Finding the Weakest Link: Adversarial Attack against Multi-Agent Communications

TL;DR

This paper presents gradient-based adversarial attack methods targeting multi-agent communication systems, demonstrating their effectiveness in various navigation environments.

Contribution

It introduces novel gradient-informed attack techniques and loss functions that improve attack success and impact on multi-agent communication systems.

Findings

Gradient-based message selection outperforms random selection in impact.

Proposed methods increase attack effectiveness in multiple scenarios.

Certain attack configurations significantly disrupt multi-agent coordination.

Abstract

Multi-agent systems rely on communication for information sharing and action coordination, which exposes a vulnerability to attacks. We investigate single-victim communication perturbation attacks against Multi-Agent Reinforcement Learning-trained systems and propose methods that use gradient information from the Jacobian to identify which messages, agent, and timesteps are most susceptible to attack and have the greatest impact on the system. We enhance these methods with two proposed adversarial loss functions that trade-off attack success for attack impact which also create more effective perturbations. We empirically demonstrate the effectiveness of our methods against two different multi-agent communication methods in navigation, PredatorPrey, and TrafficJunction environments. Our results show that our novel message selection method achieves a similar or greater impact than random message selection across almost all tested scenarios. Our victim selection, message selection, tempo, and loss functions improve attack effectiveness in half of the thirty scenarios we tested.