Resilient Distributed Fusion Filtering for Nonlinear Systems With Dynamical Bias Subject to Dual-Channel Replay Attacks

Abstract

This paper investigates the resilient distributed fusion filtering problem for nonlinear multi-sensor systems subject to dynamical bias and dual-channel replay attacks. Replay attacks are considered on both the sensor-to-local-filter and local-filter-to-fusion-filter communication channels, where current measurements and local estimates may be maliciously replaced by previously recorded data. To characterize such attacks more realistically, a unified probabilistic model involving Bernoulli stochastic variables and bounded replay delays is developed. A distributed fusion filtering scheme is then constructed, in which local filters are first designed to guarantee the exponential mean-square ultimate boundedness of the local filtering error dynamics. Sufficient conditions are established in terms of matrix inequalities, and the corresponding local filter gains are obtained by solving a constrained optimization problem. Based on the resulting local estimates, the fusion filtering performance is analyzed, and the fusion weights are further optimized through a convex optimization problem to minimize the ultimate upper bound of the fusion filtering error. Finally, a simulation example is provided to demonstrate the effectiveness and applicability of the proposed filtering method.