Hybrid-attack-resistant distributed state estimation for nonlinear complex networks with random coupling strength and sensor delays

• Published in: Journal of the Franklin Institute Vol. 361; no. 13; p. 107005
• Main Authors: Lei, Bingxin, Hu, Jun, Caballero-Águila, Raquel, Chen, Cai
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01.09.2024
• Subjects: Mean-square boundedness; Random coupling strength; Random hybrid attacks; Random sensor delays; Time-varying nonlinear complex networks; Time-varying nonlinear complex networks; Random sensor delays; Random coupling strength; Mean-square boundedness; Random hybrid attacks
• ISSN: 0016-0032
• DOI: 10.1016/j.jfranklin.2024.107005

In this paper, a recursive distributed hybrid-attack-resistant state estimation (SE) scheme is proposed for a class of time-varying nonlinear complex networks (NCNs) subject to random coupling strength (RCS) and random sensor delays (RSDs) under hybrid attacks. A hybrid-attack model is considered to characterize the random occurrence of denial-of-service (DoS) attacks and deception attacks. The objective of the problem to be solved is to develop a recursive distributed estimation method such that, in the presence of RCS, RSDs and hybrid attacks, a locally optimized upper bound (UB) on the estimation error covariance (EEC) is ensured. By employing the mathematical induction method, a UB is firstly derived on the EEC. Subsequently, the obtained UB is minimized by appropriately designing the estimator gain (EG). Furthermore, a sufficient criterion guaranteeing the exponential boundedness (EB) of SE error is elaborated in the mean square sense (MSS). Finally, simulation experiments with localization applications of multiple mobile indoor robots are conducted to illustrate the applicability of the proposed SE scheme. •A novel distributed SE scheme is given against RCS, RSDs and hybrid attacks.•The boundedness analysis with respect to SE error is given to ensure EB of SE error in the MSS.•The inversion operation of high-dimensional matrix and calculation of cross-covariance matrix between coupled nodes are reduced.