Secure Nonlinear Fusion Estimation Against FDI Attacks: The Completely Distributed Condition

This paper considers the problem of multi-sensors fusion estimation for cyber-physical systems (CPSs), where each sensor node sends a local estimate to the fusion center and a subset of the transmitted signals can potentially be under false data injection attacks. For the aforementioned problem, locating the compromised local estimates is of crucial importance for obtaining an accurate fusion state estimate of CPSs. In this case, a secure fusion estimation method is proposed based on a Gaussian mixture model based detection mechanism. Specifically, the proposed method first clusters the local state estimates autonomously and provides a belief for each sensor, based on which the local estimates can be fused accordingly. Finally, an illustrative example is employed to show the effectiveness of the proposed method.