Laura McLay, author of Punk Rock Operations Research, has an interesting research paper out on identifying risky airline passengers in order to increase security for them. It is costly (both in money and in passenger inconvenience) to subject everyone to the highest level of screening. So who should be screened, given limited screening resources? Laura worked with Sheldon Jacobson (who is frequently seen in this blog) and Alexander G. Nikolaev on this problem, and they published their work in the June 2009 issue of IIE Transactions (IIE Transactions, Volume 41, Issue 6 June 2009 , pages 575 – 591).
From the VCU Press release:
Changes in aviation security policies and operations since the Sept. 11, 2001, terrorist attacks have resulted in every passenger being treated as a potential security risk, with uniform screening of passengers and their luggage. Screening all passengers the same way is costly and inconvenient for air travelers, according to the research, published in the June 2009 issue of “IIE Transactions.”
“We set out to find a real-time screening methodology that considers both available screening resources and the necessity of being robust in assessing threat levels,” said Laura A. McLay, Ph.D., an assistant professor in the VCU Department of Statistical Sciences & Operations Research. “This paper provides methodology to quickly determine which passengers are high-risk and who is low-risk and screen them accordingly,” McLay said.
Here is the full abstract:
Designing effective aviation security systems has become a problem of national concern. Since September 11th, 2001 passenger screening systems have become an important component in the design and operation of aviation security systems. This paper introduces the Sequential Stochastic Passenger Screening Problem (SSPSP), which allows passengers to be optimally assigned (in real-time) to aviation security resources. Passengers are classified as either selectees or non-selectees, with screening procedures in place for each such class. Passengers arrive sequentially, and a prescreening system determines each passenger’s perceived risk level, which becomes known upon check-in. The objective of SSPSP is to use the passengers’ perceived risk levels to determine the optimal policy for screening passengers that maximizes the expected number of true alarms, subject to capacity and assignment constraints. SSPSP is formulated as a Markov decision process, and an optimal policy is found using dynamic programming. Several structural properties of SSPSP are derived using its relationship to knapsack and sequential assignment problems. An illustrative example is provided, which indicates that a large proportion of high-risk passengers are classified as selectees.
As the New York Times reports, the TSA is upgrading its security by requiring things like real names (not nicknames) on tickets. Perhaps operations research can offset some of the inconvenience of these “upgrades”.