Vulnerability Analysis of a Class of Nonlinear State Estimation Under Joint Deception Attacks

This paper is concerned with the vulnerability of a class of nonlinear state estimation under joint deception attacks. By decomposing the considered nonlinear system into two coupled subsystems and utilizing a decoupling method based on the subspace constructed in the kernel of the nonlinearity, we obtain the relationship between the vulnerability of nonlinear system under attacks and the stability of the two subsystems. Then the vulnerability conditions of nonlinear state estimation under joint attacks with and without resource constraints are provided, respectively, and the corresponding attack sequences are designed to degrade the estimation performance while being stealthy. Finally, a numerical example of a fixed-base inverted pendulum is provided to verify the theoretical results.