Wasserstein-Adaptive-Consensus-Based Resilient Distributed H∞ Filtering in Wireless Sensor Networks With Cyberattacks

Abstract

This article addresses the distributed state estimation problem in wireless sensor networks (WSNs) under potential threats from malicious cyberattacks. A distributed H $_{\infty }$ filtering (DHF) approach based on Wasserstein adaptive consensus is proposed to enhance estimation accuracy and resilience against such cyberattacks. First, each sensor gathers partial measurement information and updates the local prior estimate of the system by performing Krein space H $_{\infty }$ optimal filtering, followed by information exchange between neighboring sensors. Second, after obtaining local posterior estimates from neighboring sensors, a Wasserstein classification mechanism is established from the perspective of probability distribution to filter out the distributions of the state estimates from attacked sensor nodes, effectively discarding the compromised estimates. Third, an adaptive consensus scheme is designed to update the fused distribution of the state estimate for each sensor node, aiming to achieve consistent state estimation with low communication burden, which calculates the adaptive combination weights for the distributions of the state estimates from unattacked sensor nodes. Finally, the prior estimate for the next time step is predicted based on the local fused estimate at the current time step. The effectiveness of the proposed method is demonstrated by conducting simulations on specific instances of mobile target localization under various cyberattack scenarios.