Observer-Based Adaptive NN Security Control for Switched Nonlinear Systems Against DoS Attacks: An ADT Approach

Area of Science:

• Control Systems Engineering
• Cybersecurity
• Artificial Intelligence

Background:

• Switched nonlinear systems (SNSs) are susceptible to denial-of-service (DoS) attacks, disrupting sensor-controller communication.
• Standard control methods fail when controllers receive no data due to attacks.
• External disturbances and unmodeled dynamics further complicate control design for these systems.

Purpose of the Study:

• To develop a robust adaptive control algorithm for SNSs under DoS attacks.
• To design a control strategy that overcomes the limitations of traditional backstepping methods in attacked environments.
• To ensure the stability and boundedness of signals in the closed-loop system despite combined attack and switching effects.

Main Methods:

• A novel switched observer-based neural network (NN) adaptive control algorithm is proposed.
• NN adaptive observers are designed to adaptively switch based on DoS attack status.
• Dynamic surface control is employed to mitigate complexity explosion, and switching laws with average dwell time are designed using the multiple Lyapunov function method.

Main Results:

• The proposed algorithm effectively addresses the security control problem of SNSs under DoS attacks.
• The NN adaptive observers and controller ensure system operation even with sensor-controller channel disruptions.
• The designed switching laws guarantee that all signals in the closed-loop system remain bounded.

Conclusions:

• The novel NN adaptive control algorithm provides a viable solution for securing SNSs against DoS attacks.
• The integration of NN observers, dynamic surface control, and average dwell time switching laws enhances system resilience.
• The method is validated through an illustrative example, confirming its practical applicability.