New decentralized air traffic protocol uses game theory to reduce overload

Air traffic control faces growing congestion as airspace sectors operate independently with conflicting local priorities. Centralized coordination scales poorly, but fully decentralized systems often fail when sectors act self-interestedly. A new paper from researchers at UT Austin, ENAC, and École Polytechnique introduces a game-theoretic regulated protocol that bridges this gap. Their approach models each sector as a self-interested agent that adjusts flight departure times to minimize its own overload, using best-response dynamics. A tunable cooperativeness factor allows sectors to account for others' overload to varying degrees, while a minimal admissibility rule ensures local changes don't shift congestion elsewhere.

The protocol is proven to converge to a pure Nash equilibrium under a potential game structure, and under certain conditions an overload-free solution matches the global optimum. In numerical experiments using 24 hours of European flight data, the algorithm substantially reduced overload with minimal cooperation, achieving solution quality comparable to a centralized benchmark while remaining scalable. This work offers a practical path for future air traffic management systems that are both efficient and decentralized, without requiring full cooperation or heavy-handed regulation.