Hierarchical multi-agent distributed method for automated slot allocation in virtual centralized ATFM

• Main Authors: Nguyen, Nhat-Hoang P., Pham, Duc Thinh, Duong, Vu N.
• Other Authors: School of Mechanical and Aerospace Engineering
• Format: Conference or Workshop Item
• Language: English
• Published: 2025
• Subjects: Computer and Information Science; Flight scheduling; Slot allocation
• Online Access: https://hdl.handle.net/10356/182729
• Institution: Nanyang Technological University

A Virtual Centralized Air Traffic Flow Management (vATFM) is envisioned to take over the traditional institutional ATFM, particularly for regions where institutional authority is impractical, such as Asia-Pacific (APAC) region. The vATFM system relies on a central virtual node to coordinate and manage local ATFM nodes, including slot allocation for flight scheduling. Consequently, the vATFM system requires automated and distributed operations, as arranging real scheduling conferences for slot allocations at institutional ATFM centers is inoperable. We propose a distributed approach using hierarchical multi-agent design, where higher-level virtual agents coordinate and delegate tasks to lower-level virtual agents to resolve complex problems collectively. Our proposed approach utilizes a reinforcement learning mechanism where different facilities act as agents and collaborate with a virtual centralized agent to make joint decisions. This study also investigates the fairness and benefits offered by airlines. We analyze various factors, such as workloads and economic advantages, to form a set of preferences for evaluating the airlines’ satisfaction. The main objectives of the agents are to ensure fairness and resource efficiency, which are combined to construct the reward function. Experiments are conducted using simulated flight plan data in APAC region to examine the feasibility of this method in impartially addressing overcapacity issues while ensuring optimal flight schedules.