Fixed-Time Event-Triggered Impulsive Secure Synchronization of Quaternion-Valued Fuzzy Neural Networks Subject to Stochastic Cyber-Attacks

• Published in: IEEE transactions on fuzzy systems Vol. 33; no. 2; pp. 559 - 569
• Main Authors: Liu, Lirong, Bao, Haibo, Cao, Jinde
• Format: Journal Article
• Language: English
• Published: IEEE 01.02.2025
• Subjects: Artificial neural networks; Cyberattack; Delays; Event detection; Event-triggered impulsive control (ETIC); fixed-time synchronization; Fuzzy neural networks; quaternion-valued fuzzy neural networks (QVFNNs); Quaternions; Stability criteria; stochastic cyber-attacks; Stochastic processes; Synchronization; Vectors
• ISSN: 1063-6706; 1941-0034
• DOI: 10.1109/TFUZZ.2024.3485515

This article addresses the fixed-time event-triggered impulsive secure synchronization issue of a category of Takagi-Sugeno quaternion-valued fuzzy neural networks (QVFNNs) under stochastic cyber-attacks for the first time. To do this, dynamic event-triggered control, fuzzy rules, and impulsive control are combined to create a well-designed event-triggered impulsive control scheme that can significantly decline the consumption cost and communication burden. Meanwhile, in virtue of two different and independent Bernoulli random variables, the stochastic cyber-attack model involving not only denial-of-service attacks but also deception attacks is introduced. Then, some algebraic criteria are deduced to accomplish fixed-time synchronization for the QVFNNs with or without stochastic cyber-attacks through the fuzzy set theory, Lyapunov functional method, and impulsive system theory. Furthermore, the nonexistence of the Zeno phenomenon can be guaranteed. Ultimately, an illustrative instance is revealed to manifest the correctness of the theoretical outcomes and the practicability of the current strategy.