Enhancing privacy and convergence speed in self-organizing networks: A prescribed-time fully distributed neurodynamic approach with verification scheme

The execution problem of self-organizing networked systems, owing to their multi-hop and decentralized transmission structure, is commonly formulated as a networked game. In the presence of eavesdropping, unauthorized access, and the demand for rapid execution driven by limited resources and time-sensitive applications, privacy protection and rapid execution are increasingly emphasized. This paper investigates the problem of Nash equilibrium (NE) seeking under the dual challenges of long-term information leakage caused by stealthy exploratory attacks and the requirement for rapid convergence. To address these issues, a novel prescribed-time fully distributed neurodynamic approach with privacy protection against exploratory attacks is proposed. Combining a double-layer adaptive approach with the prescribed-time technique, the approach achieves the prescribed-time NE seeking and effectively prevents the overgrowth of adaptive control gain. The core of privacy protection lies in an initiative switching mechanism that proactively disconnects potential data observation channels, thereby protecting private information without the need for attack detection. The convergence of the approach is shown via Lyapunov stability theory. Moreover, to evaluate the outcome of privacy protection, a verification scheme is constructed using a virtual attacker cluster. Applied to a self-organizing network of autonomous vehicles, simulation examples demonstrate the effectiveness of the proposed approach in both privacy protection and convergence performance.