Characterizing Trust and Resilience in Distributed Consensus for Cyberphysical Systems

Описание: Title: Characterizing Trust and Resilience in Distributed Consensus for Cyberphysical Systems
Speaker: Angelia Nedich
Abstract: This talk considers the problem of resilient consensus in presence of malicious agents in the setting where the agents can learn the trustworthiness of their neighbors over time. Specifically, we propose and analyze the model where stochastic values of trust between agents are available. We provide a unified mathematical framework to characterize convergence, deviation of the achieved consensus from the true consensus value, and the expected convergence rate. We show that under certain conditions on the stochastic trust values and consensus protocol: 1) almost sure convergence to a common limit value is possible even when malicious agents constitute more than half of the network connectivity, 2) the deviation of the converged limit, from the case where there is no attack, i.e., the true consensus value, can be bounded with the probability that approaches 1 exponentially, and 3) correct classification of malicious and legitimate agents can be attained in a finite time almost surely.