ECE professor Constantine Caramanis and colleagues at MIT are working on a air traffic system which would track rapidly changing conditions over airports. There is currently no unified decision-making framework for air traffic flow optimization, said Dr. Caramanis.

The complicated nature of the process, and the need to make quick adjustments when changes occur, will best be addressed with a mathematical model that combines theories and calculations from probability, statistics, optimization modeling, economics and game theory.

A system for streamlining decisions about airplane takeoffs and landings that adjusts to new input on the fly is under development by a University of Texas at Austin engineer and Massachusetts colleagues.

The mathematical optimization model Assistant Professor Constantine Caramanis and colleagues will develop in the next five years will provide a decision-making system that rapidly adapts its flight recommendations without human input based on thousands of changing variables.

There is currently no unified decision-making framework for air traffic flow optimization, said Caramanis, an electrical engineer at the university who will collaborate with lead researcher Cynthia Barnhart and other colleagues from Massachusetts Institute of Technology (MIT) to develop the computer model using a $2 million grant from the National Science Foundation. The complicated nature of the process, and the need to make quick adjustments when changes occur, will best be addressed with a mathematical model that combines theories and calculations from probability, statistics, optimization modeling, economics and game theory, he said.

The Federal Aviation Administration (FAA) provides each airline with a set limit of planes that can take off and land during any given timeframe. These slot decisions are based on estimates of what will optimize air traffic flow, taking into consideration imperfect weather predictions, the changing mix of flights airlines wants to move, and other variables for the thousands of flights that crisscross U.S. skies daily.

The airlines then choose which flights to fill their slots with, and have differing priorities that can also reduce the overall flow of air traffic.

Caramanis will use his expertise in developing optimization models that involve uncertainty as he and his colleagues apply the National Science Foundation funding to inform air traffic decisions. As a starting point, the researchers will evaluate recent flight-related information to develop benchmarks for progress, and understand the potential of the models they will develop.

Computer crashes and other factors that can impact how soon the FAA receives information will not be considered. Instead, the researchers will focus on air traffic flow-related information once it arrives. For his part in the research, Caramanis received $600,000 of the foundation funds to help create the model that will adapt to new input on its own (called autonomous reconfigurability).

While developing the air traffic optimization model, the researchers will also consider new ways to lessen delays and flight cancellations. For example, they will consider the possibility of allowing airlines to barter for slots when one airline can't get a flight off the ground and others could do so.

The idea is to have an overarching optimization model that allows balance and flexibility to the decisions being made so that we can successfully exploit whatever slack in the system we can, Caramanis said.

The grant is part of the National Science Foundation's Emerging Frontiers in Research and Innovation program.