Expectation-Maximization Based Defense Mechanism for Distributed Model Predictive Control

TL;DR

This paper introduces an Expectation-Maximization based algorithm to defend against communication attacks in distributed model predictive control, improving robustness in large-scale, decentralized systems.

Contribution

It presents a novel EM-based method specifically designed to mitigate attack effects in distributed control systems, enhancing security and reliability.

Findings

Effective mitigation of communication attacks demonstrated in temperature control example

Improved robustness of distributed predictive control under adversarial conditions

Algorithm shows promise for secure large-scale system management

Abstract

Controlling large-scale systems sometimes requires decentralized computation. Communication among agents is crucial to achieving consensus and optimal global behavior. These negotiation mechanisms are sensitive to attacks on those exchanges. This paper proposes an algorithm based on Expectation Maximization to mitigate the effects of attacks in a resource allocation based distributed model predictive control. The performance is assessed through an academic example of the temperature control of multiple rooms under input power constraints.