Resilient Quantized Consensus in Adversarial Environment Using Trusted Neighbors

This paper addresses the challenge of achieving resilient quantized consensus in multiagent systems operating within adversarial environments. Existing algorithms often rely on an assumption of the maximum adversarial agents in the multiagent systems or neighbors which may be impossible to appropriately estimate in practice. To address this limitation, we propose a novel filtering method that dynamically maintains a trusted neighbor set to identify and filter out suspicious states without requiring knowledge of the maximum adversaries. Additionally, we introduce a distributed resilient quantized control protocol tailored for regular agents, ensuring consensus despite limited communication and memory constraints that enforce agents to adopt integer values. We analyze a graph-theoretic condition that is both necessary and sufficient for our proposed control protocol to succeed. Finally, we validate our theoretical findings through simulations.