Quantized sampled-data control for synchronization of switching stochastic neural networks with time-varying delays and random packet dropouts: Application to secure communications

This paper investigates the synchronization problem of stochastic neural networks (SNNs) characterized by Markovian switching parameters and time-varying delays (TVDs), within the framework of physical systems. In such systems, signal transmission between neurons is often influenced by intrinsic challenges such as time delays, external disturbances and uncertain parameters, all of which may degrade system performance and communication efficiency. To address these issues, a quantized memory sampled-data control (QMSDC) approach is proposed to enhance the robustness and physical realism of neural network models. A novel looped-type Lyapunov functional (LTLF) is constructed, integrating both sampling information and the effects of quantized communication and transmission delays. Based on Jensen’s integral inequality, sufficient synchronization conditions are derived under probabilistic packet loss and are formulated as linear matrix inequalities (LMIs) to facilitate numerical verification. The effectiveness of the proposed method is demonstrated through numerical simulations, which confirm improved synchronization performance in delayed SNNs operating in physical environments. Furthermore, the chaotic dynamics inherent to such delayed systems are exploited to develop a secure communication scheme. The security aspect is validated through statistical analyses on standard benchmark images, verifying the protocol’s capability in ensuring reliable and confidential information transmission in physical neural systems.