Balancing air traffic demand with capacity is critical to ensuring the smooth and orderly flow of flights. With rising passenger numbers and increased flight volumes, new innovative technologies will be needed to help flexibly manage ever more congested airspace. To ease congestion, simple swaps between two flights can currently take place. Such exchanges help airlines to prioritise certain flights, according to flight cost structures, to minimise costs and reduce passenger delays. For airlines, the prioritization – and swapping – of these flight slots is therefore a critical cost issue.
A challenge here is that flight cost structures, which influence how airlines prioritise flights and may vary for any number of reasons, are highly confidential. This is not information that airlines necessarily want to share with others and has hampered slot swapping between different airlines in the past.
The SlotMachine project sought to harness the power of artificial intelligence (AI) technology to create a semi-automated flight allocation swap platform, based on the cost structure priorities of different airlines. Airspace users can securely input their flight preferences, together with supporting data, via a graphic interface. This data is sent to the system’s heuristic optimiser, where genetic algorithms are applied to the combined data, to find incrementally improved solutions in terms of flight prioritisation.
The prototype builds on the principle of separating the search for optimal flight lists from the evaluation of the flight lists. The evaluation is conducted by the Privacy Engine, which employs MPC to evaluate the solutions. MPC nodes external to the SlotMachine system conduct the necessary computations. No single MPC node knows the full inputs. Different parties, e.g. different AUs, may host the MPC nodes. The employed architecture could also potentially be employed for other optimization problems. The SlotMachine project has the following main objectives:
- Delay and cost optimization. The delay and, more importantly, the delay costs should be reduced by allowing airspace users to prioritize flights and exchange ATFM slots in case of reduced capacity in the air traffic network following a regulation. The success of the optimization can be measured either by decreased costs or, conversely, by increased utility, if the preferences of airspace users are stated in terms of economic utility of a slot for a flight. A maximization of utility leads to a minimization of the costs.
- Equity and fairness over time. Airspace users giving up favourable slots should be compensated. We propose market mechanisms based on credits. Prioritization of a flight requires the airspace user to spend credits, accepting additional delay leads to the airspace user receiving credits which can be spent to prioritize flights in the future.
