Fixed-Time Consensus of Multi-Agent Systems in Asymmetric Signed Networks With Application to UAV Swarms

A fixed-time consensus protocol for multi-agent systems in asymmetric signed networks is proposed by integrating evolutionary game theory with a dual-mechanism incentive approach. Three key challenges are addressed through novel mechanisms. First, asymmetric interactions among agents in signed networks are characterized through a modified evolutionary game framework that captures both cooperative and competitive relationships. Second, an adaptive consensus protocol is proposed, enabling the system to achieve fixed-time convergence despite time-varying network topologies. Third, a DQN-optimized hierarchical external incentive mechanism is designed to accelerate consensus. Cooperation among agents is effectively promoted by combining structural importance-based stratification with dynamic reward allocation. Incentive parameters are dynamically adjusted based on network states and agent behaviors through the DQN optimization framework, resulting in significant convergence efficiency improvements. Convergence is guaranteed precisely within a fixed time bound through a comprehensive Lyapunov-based stability analysis, regardless of initial conditions. The theoretical fixed-time convergence is verified through collaborative search mission simulations with UAV swarms, demonstrating the ability to achieve consensus in asymmetric signed networks with both cooperative and antagonistic relationships.