Trust-Based Sequential Monotonic Concession Protocol for Two-Player Task Negotiation

We propose a trust-based bilateral negotiation mechanism for task allocation problems under incomplete information, where a deceptive agent may disseminate false information to manipulate the outcome. Although numerous techniques have been explored to identify and reject such malicious agents, these approaches are often impractical due to insufficient prior knowledge, distributed implementation, or dependency on external entities. In contrast, the proposed mechanism enables agents to autonomously mitigate deception through trust evolution and distributed decision-making while acknowledging possible deception. The proposed mechanism integrates trust and cooperative utility into the Monotonic Concession Protocol, where the extended Zeuthen strategy forms a Nash equilibrium. A trust-weighted utility function is introduced, allowing agents to adaptively adjust their negotiation behavior based on the opponent’s behavior. We provide theoretical analysis of the mechanism, including Nash equilibrium, possible deception, and reachability. During negotiation, agents iteratively update trust, which governs both their degree of cooperation and information disclosure. A higher trust level encourages collaboration and gradual information sharing, while lower trust promotes selfish behavior and limited information exposure. Extensive Monte Carlo simulations demonstrate that the proposed mechanism effectively mitigates deception and enhances negotiation outcomes across various deception scenarios. Note to Practitioners—This research is motivated by the task allocation problem, specifically focusing on the cost functions that can be directly mapped into the traveling salesman problem. All simulation examples presented in this paper employ dynamic programming to compute exact cost functions and identify the optimal solution. However, given the combinatorial nature of the problem, this technique may prove impractical within feasible time constraints, especially when dealing with a large number of tasks or a complex environment with multiple obstacles. In such cases, a heuristic approach could serve as an alternative that alleviates computational burden.