Secure Distributed Model Predictive Control for Heterogeneous UAV-UGV Formation Under DoS Attacks

Abstract

This study addresses the secure distributed model predictive control (SDMPC) challenge for the heterogeneous UAV-UGV formation system under malicious denial-of-service (DoS) attacks, utilizing a nonlinear discrete-time model to represent system dynamics. It examines the scenario where DoS attacks obstruct communication between neighboring agents. A novel neighbor output prediction strategy is introduced to mitigate the impact of DoS attacks. Upon detecting a DoS attack, subsystems affected by the compromised channel predict the output sequences of their upstream counterparts, updating these predictions at each time step based on receiver buffer contents and attack duration. Subsequently, a cost function incorporating the predicted output sequences and a terminal constraint tailored to DoS conditions is formulated to maintain system stability during attacks. The analysis thoroughly explores recursive feasibility and input-to-state practical stability (ISpS). Comparative tests underscore the proposed SDMPC algorithm's effectiveness and enhanced security in maintaining stability amid DoS attacks.