Event-Triggered Secure Fusion Estimation with Watermarking Against Linear Man-in-the-Middle Attacks in Multirate Multisensor Systems

In this paper, event-triggered secure fusion estimation is investigated for a multirate multisensor system under linear man-in-the-middle (MITM) attacks and correlated heavy-tailed noises. A dynamic event-triggered mechanism is proposed to reduce measurement information transmission over communication networks with linear MITM attacks. A watermarking strategy is designed to address impact of linear MITM attacks by fully recovering measurement information. An event-triggered local estimator is designed to obtain local estimates under correlated heavy-tailed noises and event-triggered measurements. A distributed fusion estimator is presented to improve estimation accuracy based on sequential fast covariance intersection fusion technique. Sufficient conditions are given to ensure boundedness on estimation error scale matrices of the event-triggered local estimator and the distributed fusion estimator. Simulation and experimental results are given to illustrate validity of the event-triggered secure fusion estimation. Note to Practitioners - This paper addresses event-triggered secure fusion estimation problem in practical engineering scenarios, such as autonomous mobile robots, smart manufacturing systems, industrial internet of things and so on. In these scenarios, multiple sensors often measure data at different sampling frequencies, which results in challenges for fusion estimation. Measurement information may occasionally exhibit outliers, which degrades estimation accuracy under Gaussian distribution assumption. Transmitting all measurement information overloads communication networks, increases energy consumption and reduces bandwidth efficiency. In wireless communication networks, cyber-attacks tampers with sensor data, which decreases estimation accuracy and causes system failures. To address these issues, a dynamic event-triggered mechanism is designed to reduce measurement information transmission over wireless communication networks. A watermarking strategy is proposed to fully recover measurement information destroyed by linear MITM attacks for state estimation. Furthermore, event-triggered secure fusion estimation with watermarking is investigated for multirate multisensor systems subject to linear MITM attacks and correlated heavy-tailed noises. Simulation and experiment results illustrate effectiveness of the proposed algorithm. Future work will focus on adaptive fusion estimation resilient to multiple cyber-attacks or sensor failures under varying network conditions.