Distributed secure state estimation of heterogeneous multi-agent system under unknown exogenous attack

This paper investigates the issues of distributed information fusion and secure state estimation in heterogeneous multi-agent system. In these systems, agents are characterized by distinct state and observation equations and can exchange observations and state information through a communication network, introducing significant challenges for information fusion. Based on Bayesian theory, the proposed method efficiently integrates heterogeneous information to provide minimum mean square error (MMSE) state estimates for each agent and the convergence of algorithm is proved. Furthermore, if systems suffers False data injection(FDI) attack, an attack-decoupling strategy is developed to mitigate the impact of exogenous attacks, ensuring the security and accuracy of state estimation under adversarial conditions. Finally, simulations demonstrate that the proposed filter can achieve accurate estimation, and the secure algorithm is effective.