Fixed-time synchronization of delayed multiple inertial neural network with reaction-diffusion terms under...

Area of Science:

• Control Theory
• Neural Networks
• Cybersecurity

Background:

• Delayed multiple inertial neural networks (MINNs) are susceptible to cyber-physical attacks (CPA).
• Designing secure control laws for MINNS with reaction-diffusion (RD) terms under CPA is challenging.
• Ensuring fixed-time synchronization (FXTS) is crucial for network stability and data integrity.

Purpose of the Study:

• To investigate the fixed-time synchronization (FXTS) problem for delayed multiple inertial neural networks (MINNs) under cyber-physical attacks (CPA).
• To develop innovative and practical criteria for achieving FXTS using fixed-time stability theory.
• To propose a robust security control strategy to counteract CPA and ensure network synchronization.

Main Methods:

• Utilizing fixed-time stability theory to derive FXTS criteria.
• Developing a distributed control strategy for delayed MINNs with reaction-diffusion (RD) terms.
• Employing Lyapunov functions and M-matrix properties to analyze CPA effects.
• Designing a security control law to guarantee FXTS.

Main Results:

• Innovative criteria for achieving FXTS in delayed MINNs against CPA were derived.
• A distributed control strategy was successfully introduced to attain FXTS.
• The security framework and control algorithm enable parameter selection for feedback gain and coupling strength.
• A numerical model validated the theoretical findings.

Conclusions:

• The proposed security control strategy effectively guarantees FXTS for delayed MINNs under CPA.
• The study provides a robust framework for securing neural network systems, particularly in sensitive applications like healthcare.
• The developed multi-image encryption algorithm enhances data integrity in electronic healthcare systems.