Adaptive PD-Like Event-Triggered Secure Synchronization Control for Inertial Neural Networks and Signal Encryption...

Area of Science:

• Control Theory
• Cybersecurity
• Artificial Intelligence

Background:

• Markovian jumping delayed inertial neural networks (INNs) are crucial for signal processing.
• Ensuring secure synchronization under hybrid attacks (Denial-of-Service and Deception Attacks) is a significant challenge.
• Existing methods may not efficiently handle redundant data sampling in complex network environments.

Purpose of the Study:

• To investigate the exponential secure synchronization of Markovian jumping delayed INNs under hybrid attacks.
• To propose a novel adaptive proportional-derivative (PD)-like event-triggered mechanism (APDETM) for efficient data sampling.
• To design event-triggered output feedback controllers for secure synchronization control.

Main Methods:

• Development of an adaptive proportional-derivative (PD)-like event-triggered mechanism (APDETM) considering state variations.
• Establishment of INNs with generally uncertain semi-Markovian (GUSM) jumping parameters under hybrid attacks.
• Design of event-triggered output feedback controllers for secure synchronization.

Main Results:

• The proposed APDETM effectively filters redundant sampling data while maintaining system performance.
• Secure synchronization conditions were derived for INNs with GUSM parameters under hybrid attacks.
• The developed control strategies demonstrated effectiveness in numerical simulations and audio encryption.

Conclusions:

• The proposed event-triggered control approach ensures exponential secure synchronization for delayed INNs under hybrid attacks.
• The APDETM offers an efficient solution for data sampling in networked systems.
• The application in signal encryption, particularly audio encryption, highlights the practical utility of the method.