Recursive Resilient State Estimation for Nonlinear Stochastic Complex Networks With Energy Harvesting Sensors Under Deception Attacks

Abstract

This paper deals with a resilient estimation problem for certain type of time-varying complex networks of energy harvesting sensors that are vulnerable to deception attacks. Measurement signals of the underlying complex network, as measured by energy harvesting sensors, are only given to a remote estimator when the energy level is adequate to offset the energy consumption, which is at risk of deception attacks during network transmission. The deception attacks under consideration, are depicted as events occurring randomly, governed by a Bernoulli sequence. To meet the desired estimation performance, a resilient scheme is developed that addresses the side effects of random perturbations of the estimator gain when it comes to the implementation. The primary objective is to devise a resilient algorithm that can simultaneously manage energy harvesting sensors, deception attacks, and gain perturbations of the state estimator. Initially, the upper bound of the obtained error covariance is determined by making use of induction and intensive stochastic techniques. The necessary estimator gains are then identified recursively to prudently minimize this acquired bound. An illustrative example is presented ultimately to demonstrate this scheme's efficacy.