Every time a traveler’s phone pings a cell tower, it leaves a tiny digital footprint.
A new international study co-authored by a Texas A&M College of Agriculture and Life Sciences researcher shows those footprints, multiplied by millions, can help airports predict passenger surges before they happen, and cut the congestion and emissions that come with them.
The study, published in Business Strategy and the Environment, used more than 9 million anonymized mobile network signals collected around Lisbon Airport in Portugal to build a forecasting model that outperformed standard prediction methods by double-digit margins, cutting forecast errors by up to 24%.
People are also reading…
Babak Taheri, professor and associate department head of graduate programs in the Arch H. Aplin III ’80 Department of Hospitality, Hotel Management and Tourism, was part of the research team, along with faculty from Molde University College in Norway and Inov Inesc in Portugal.
“Lisbon was a strong test case because it brings together many of the challenges airport managers deal with every day,” Taheri said. “It has high passenger volumes, strong seasonal peaks, a mix of domestic and international travelers and real pressure on capacity, congestion and ground transportation.”
From data to decisions
The team tracked anonymized, aggregated signals from mobile devices across 119 grid cells covering the airport’s footprint over a full year, then fed the patterns into a forecasting model built on Prophet, a time-series tool suited to capturing seasonal swings and holiday effects. Any grid-and-time slice with fewer than 10 devices was excluded, and no individual users could be identified.
What sets the study apart is what happens after the forecast. The research team translated those projections directly into potential operational decisions. Their forecast showed how the Lisbon Airport could use the information to determine how many security lanes to open, how many staff members to schedule and how much to shorten the intervals between metro trains.
Under the scenarios modeled, better-coordinated staffing and transit schedules could reduce ground-transport congestion and associated emissions by 14-26% during peak travel periods.
“Telling an airport manager that passenger demand may increase by a certain percentage is only partly useful,” Taheri said. “Telling them what that could mean for staffing or the number of security lanes makes the information much more actionable.”
The model is not perfect. For example, it underestimated actual passenger counts during a June 2023 spike, a gap that Taheri said underscores why forecasts should inform planning ranges rather than serve as a single guaranteed number.
“There will always be days or periods when actual demand moves outside expectations,” he said.
International passenger flows also proved harder to predict than domestic ones, likely because they are shaped by holiday calendars, economic conditions and other factors across multiple countries. This finding points to combining mobile data with flight schedules and weather data for sharper forecasts, Taheri said.
A model for other hubs
With global air travel projected to grow by the billions over the next two decades, the researchers argue mobile network data is an underused asset for sustainability-minded airport management.
Taheri cautioned, however, that the approach is not plug-and-play.
“Every airport has its own passenger mix, terminal layout, transfer traffic and seasonal patterns, so the model would need to be trained and calibrated using local data,” he said.
The staffing and security thresholds outlined in the study have not yet been tested in live operations.
“The current study has not yet been deployed as a live airport operating system,” Taheri said. “What we have demonstrated is that the data and forecasting can be translated into decisions airport managers recognize and act on. The logical next step is an industry pilot where we test those decisions in real time, measure what works and refine the system with airport operators.”
If validated in real-world operations, the approach could give airports a new tool to anticipate pressure points before terminals and transportation networks become congested.
