Agent-based Optimization of Distributed Air Traffic Management

In the transition from current Air Traffic Management (ATM) practices to the envisioned Distributed Air-Ground Traffic Management (DAG-TM) environment, inter-aircraft negotiation protocols will prove to be a key driver in shaping the balance between traffic safety, flow rates, and individual carrier operating objectives. We propose to develop an agent-based simulation that will serve as a tool to explore these issues, by upgrading our existing in-house ATM simulation to include one or more state-of-the-art human decision-making models best suited for the ATM environment. We first propose to survey several modeling approaches that have potential in the ATM context, and then evaluate these models across a number of decision-making criteria, including: social welfare, individual rationality, stability, computational efficiency, scalability, and convergence. We will then implement one or more of the most promising decision-making models within an existing agent-based architecture, evaluate the representational fidelity of these agent-based models, and assess their utility in predicting the impact of different protocols on overall air traffic operations. We will also assess usability from the standpoint of the system designer in formulating effective negotiation guidelines for future ATM practices.