Adaptive NN-based security control for switched uncertain systems under DoS and deception attacks

This paper investigates security control strategies for uncertain switched systems under Denial-of-Service (DoS) attacks and deception attacks, where the two attack types occur alternately. DoS attacks are mitigated by zeroing the input signal, while deception attacks are assumed to satisfy a more restrictive condition. To address the uncertain function in the system dynamics, a radial basis function neural network (RBFNN) is employed. An adaptive law is formulated to estimate the RBFNN weight vector, enabling approximation of the uncertainty terms. Based on this framework, a stability criterion is established to ensure that the uncertain switched system remains practically stable under both DoS and deception attacks. Finally, a simulation on a turbofan engine is conducted to validate the effectiveness of the proposed approach.