Airport Security, AI Among DHS Lab's Tech Innovation Projects

Airport Security, AI Among DHS Lab's Tech Innovation Projects

Arizona State University is applying analytics and other tools to help the agency predict and respond to emergencies.

A university under the Department of Homeland Security's center of excellence program is helping the agency innovate technologies around various mission areas including long security lines at airports, automation in procurement, and detection of smuggling routes.

Unlike other centers, the Center for Accelerating Operational Efficiency (CAOE) led by Arizona State University supports research and development efforts for the agency's IT functions across more than one area.

For example, the center is currently testing a project with DHS Science and Technology Directorate that will help the Transportation Security Administration (TSA) better allocate resources and manpower to handle large queues of passengers at airport checkpoints.

“The idea here is they're building technology to predict arrival times of passengers at the airport then using this to feed into TSA officers' scheduling so we have the right number of officers scheduled to the right gates at the right time to improve throughput,” the center's director, Ross Maciejewski, told GovernmentCIO Media & Research. “Our initial project was dynamic resource allocation. Once we understand the demand, how do we dynamically adjust staffing models of TSA officers to improve wait times?"

The center turned the TSA project into an educational opportunity for minority institutions, Maciejewski added.

“What was really exciting from the whole resource ecosystem here was this involved undergrads collecting data at the airport counting passengers in line, we extended this to a whole summer training program at minority institutes, then giving undergrad students there training in data analytics operations research, taking classes three hours a day then data three hours a day,” he said.

The CAOE is also working on a project totally unrelated to the TSA — working with the Procurement Innovation Lab (PIL) at DHS headquarters to streamline the Contractor Performance Assessment Reporting System (CPARS) with artificial intelligence and machine learning.

“We looked directly with PIL where they're trying to figure out how to improve the culture of procurement and process contracts in a more efficient manner,” Maciejewski said. “There's a lot of overhead. If you can save man hours, you can save a lot of money. We had all these records of past contracts and had to manually sort through them to make determinations to award a new contract. Computers are really good at helping with these sort of things, so we wanted to see if there was some AI or ML to guide them in making determinations and support workflow to make them more efficient.”

PIL contracted different companies to try this method, then CAOE helped evaluate different solutions. Led by professor Thomas Cole, the group is “exploring a competing values framework assessment tool to evaluate their process control, innovation, human resources and outcomes.”

“It's really exciting to see how the PIL group has taken their large treasure trove of data to improve operational efficiencies,” Maciejewski said.

That’s just the tip of the iceberg for this center of excellence.

Maciejewski said the CAOE has only been around for four years, but two projects are already out for field testing: one is the TSA project, which currently testing at the Phoenix Sky Harbor Airport in Phoenix, Arizona. The other is the Simulation, Analysis and Modeling for Border Apprehension and Security, or SAMBAS, project.

“They're developing this tool looking at previous apprehensions from [Customs and Border Protection] with respect to smuggling, then looking at border geography and identifying most likely smuggling routes," Maciejewski said. "The goal is to deploy this at the CBP Tuscan Sector in the coming months.”

The goal of all the CAOE’s projects is to use qualitative analysis to improve resource allocation and cut costs for DHS.

“The bottom line is improving operational efficiency at DHS,” Maciejewski said. “How can we take this treasure trove of data that's being captured every day, and growing, and combine it with qualitative analytics methods like AI, ML, combine with operational techniques and then report this in terms of economic analysis and risk."

When the COVID-19 pandemic hit the U.S. last year, the CAOE stepped in to assist (similar to other centers of excellence), influencing the city of Austin, Texas’ response to the virus.

“We've been lucky that our center has always been focused on projects directly related to the operations of DHS,” Maciejewski said. “We had projects already related to disaster response. One was about how can we optimally redeploy resources in the case of a disaster like a hurricane landing. We can transition a lot of these modeling activities to look at COVID. Where should we be deploying resources, allocating different things for response. That was very successful.”

The CAOE also used economic modeling to determine how COVID-19 and pandemic-focused policies would impact different sectors of the U.S. economy.

“They were using a quarterly equilibrium model to look at the U.S. economy and associated policy responses,” Maciejewski said. “They had applied this to the meatpacking industry to see if the current safety measures would have a livescale impact on the economy. They were trying to find out which sectors would be the most impacted and see where aid would need to be deployed to promote a resilient recovery of these sectors.”

Looking ahead, the researchers are looking into subject areas like AI, ethics and algorithmic fairness due to DHS’ increasing reliance on biometric technology and facial recognition.

“We have a new project coming out of [The College of] William & Mary where they're trying to see whether they can use deep learning of satellite imagery to predict migration flows along the border," Maciejewski said. "That's a high risk and high reward, but if it can be done, that would be a great tool to predict migration patterns.”

Standard