Optimal Containment Control for Stochastic Multiagent Systems via Simplified ADP Under Secure Communication.

Area of Science:

• Control Systems Engineering
• Cybersecurity
• Artificial Intelligence

Background:

• Multiagent systems (MASs) face challenges in secure communication and coordinated control.
• Optimal Containment Control (OCC) is crucial for MASs, requiring followers to stay within a region defined by leaders.
• Stochasticity and security concerns complicate the design of effective OCC strategies.

Purpose of the Study:

• To develop a novel optimal containment control (OCC) strategy for nonlinear stochastic multiagent systems (MASs) with secure communication.
• To ensure followers converge to the convex hull of leaders while minimizing control cost.
• To address the challenges posed by stochastic dynamics and secure information exchange.

Main Methods:

• A novel OCC strategy is designed incorporating an encryption/decryption mechanism for secure communication.
• A simplified adaptive dynamic programming (ADP) framework is introduced to solve the stochastic Hamilton-Jacobi-Bellman (HJB) equation.
• A single critic network weights tuning rule is developed using the experience replay technique (ERT).

Main Results:

• The proposed OCC strategy ensures followers converge to the convex hull of leaders.
• Secure communication is maintained through encryption and decryption, preserving data integrity.
• The adaptive dynamic programming approach with ERT effectively solves the HJB equation, relaxing traditional requirements.

Conclusions:

• The developed OCC strategy is effective for nonlinear stochastic multiagent systems under secure communication.
• The integration of ADP and ERT provides a robust framework for solving complex control problems.
• The theoretical analysis confirms the uniform ultimate boundedness of the closed-loop system, validating the approach.