In the context of increasingly complex and dynamic multi-agent systems, ensuring secure and effective coordination among agents poses significant challenges. This study proposes a coordination control method based on evolutionary game theory under a zero-trust architecture. Agents are situated in a structured population on a static undirected network and engage in division-of-labor games with their neighbors, selecting between two complementary roles. A novel comprehensive evaluation mechanism is introduced, integrating both fitness and dynamically updated trust. The trust model incorporates historical cooperation success rates with exponential decay and real-time verification outcomes, aligning with the principles of continuous authentication and dynamic authorization in zero-trust security. Through theoretical analysis and weak selection approximation, we derive the system’s evolutionary dynamic equations and obtain explicit expressions for equilibrium conditions and equilibrium points. The results demonstrate that our model not only facilitates stable strategy evolution but also adapts effectively to dynamic environments and heterogeneous agent behaviors. This work provides a feasible theoretical framework and practical reference for secure and adaptive coordination in multi-agent systems under zero-trust constraints.

Zero-trust architecture, Multi-agent systems, Evolutionary game theory, Dynamic trust evaluation