A resilient consensus algorithm with inputs for the distributed monitoring of cyber-physical systems

Recent advancements in multi-agent systems (MASs) have led to the development of numerous algorithms for achieving specific objectives, such as consensus. However, security remains a major challenge in MAS consensus, particularly addressing the adversarial behavior of malicious agents. This paper explores the extension of Mean-Subsequence-Reduced (MSR) algorithm-type mechanisms for resilient dynamic consensus in the presence of input reference signals. We provide necessary and sufficient conditions for resilient dynamic consensus without relying on the presence of trusted agents. Additionally, we experimentally validate the proposed algorithm and related conditions over a small cyber-physical system used for temperature monitoring. Furthermore, we propose and experimentally validate a fault-detection and recovery algorithm to achieve a resilient dynamic average consensus of regular agents.