Dec 22, 2011 - Claremont, Calif. - Susan Martonosi, associate professor of mathematics, recently appeared on “The Academic Minute,” a radio program that features professors from the world’s top institutions.

The segment, which aired Dec. 21, shared how Martonosi developed a mathematical model to reduce the wait times in airport security lines.

Martonosi examined the benefits of queue control after a meeting in 2004 with Transportation Security Administration (TSA) officials at San Francisco International Airport. There, she observed the methods TSA officials used to manage the flow of people through airport security checkpoints.

When one line would grow longer than another, they would shift a team of inspectors to that terminal to reduce waiting times. Noting that TSA officials were making these real-time decisions using observed queue lengths, knowledge of flight schedules and past experience, Martonosi wondered if a mathematical model could improve the process and reduce wait times.

She used an approximate dynamic programming framework to model a queuing system consisting of two parallel queues, and tested the model on both synthetic and security checkpoint data from Boston Logan International Airport. At every decision point (about every 30 minutes), the model determined whether or not to switch one or more servers from one queue to the other to minimize traveler wait times. The tradeoff was that when a switch occurred, the server was off-duty for a short time to reflect the time required to walk from one queue to the other.

The results revealed that in situations where customer arrival times fluctuate significantly, dynamic reallocation of servers to queues could reduce customer wait times. However, the optimal decision of when to reallocate servers did not obey simple rules that a manager could use on-the-fly, and using seemingly common sense rules could increase waiting times.

Most airports can meet the “industry standard” goal of getting passengers through security in 10 minutes or less, Martonosi said, although it may sometimes feel longer.

“They do a pretty good job of anticipating and adapting to high volumes of travelers, including holiday travel times,” she said. “When they may run into trouble is when outside factors, such as weather delays that often happen during the winter holidays and sudden gate changes cause last-minute disruptions.”

That, she said, is when using a mathematical model may come in handy.

Listen to The Academic Minute segment.