Guaranteed performance control for delayed Markov jump neural networks with output quantization and data-injection attacks

• Published in: International journal of machine learning and cybernetics Vol. 16; no. 1; pp. 173 - 188
• Main Authors: He, Lanlan, Zhang, Xiaoqing, Jiang, Taiping, Tang, Chaoying
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.01.2025; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Communication; Complex Systems; Computational Intelligence; Control; Control systems design; Controllers; Design factors; Design optimization; Engineering; Feedback control; Markov analysis; Markov chains; Mechatronics; Neural networks; Optimization algorithms; Original Article; Output feedback; Pattern Recognition; Performance indices; Robotics; Systems Biology; Upper bounds; Output quantization; Data-injection attack; Markov jump neural network; Output-feedback control
• ISSN: 1868-8071; 1868-808X
• DOI: 10.1007/s13042-024-02195-3

This paper considers guaranteed performance control for delayed Markov jump neural networks (DMJNNs) under output quantization and data-injection attacks. The objective is to design an asynchronous output-feedback controller (OFC) that takes into account both quantization and attacks to achieve stochastic stability and ensure the boundedness of a predefined performance index. An exponential hidden Markov model is employed to represent the asynchrony between the modes of the OFC and the DMJNN. A sufficient condition for the desired performance is presented using free-weight matrix and Lyapunov–Krasovskii functional methods, integral inequalities, and Dynkin’s formula. Two distinct controller design approaches are proposed, depending on whether the coefficient matrix of the control input is a unit matrix while considering factors related to attacks and quantization. Optimization algorithms are developed based on the proposed controller design approaches, allowing for the determination of the minimum upper bound of the predefined performance index and the accompanying controller gains. Finally, a simulation example is provided to illustrate the applicability and effectiveness of the optimization algorithms developed.