AT&T introduces Geo Modeler AI system to boost network reliability during extreme weather

Extreme weather events can put stress on mobile networks, causing people to lose connectivity at the worst possible time — in the midst of natural disaster.

So, to improve network reliability and support first responder communication during emergencies, AT&T has implemented Geo Modeler, a new generative AI system that dynamically adjusts its mobile network. AT&T shared the news in an exclusive announcement to Mashable, and company leaders said the AT&T Geo Modeler system is now in effect nationwide.

"If we have a disaster, an outage or if cell towers are undergoing maintenance, Geo Modeler in concert with our network automation autonomously compensates to minimize customer impact and maintain connectivity for emergency personnel," said Raj Savoor, VP of Network Analytics and Automation at AT&T.

Using generative AI, Geo Modeler will predict strain on the AT&T mobile network, allowing the company to be proactive. That could mean adjusting the tilt of a radio antenna, boosting power to a specific signal, or making recommendations on where to deploy extra equipment.

Savoor said that AT&T has been using an early version of Geo Modeler in the field this past year, including during the deadly flash floods that hit Central Texas in July.

Geo Modeler is just the latest example of scientists using AI to track, predict, and respond to extreme weather events, which are becoming more common as the climate warms.

"Imagine utilizing AI to adapt to unexpected changes in our network infrastructure, remotely resolving coverage gaps or other challenges in near real-time," Savoor wrote in an AT&T blog post that will be published later today. "This is now a reality and being delivered by the AT&T Geo Modeler."

So, how does Geo Modeler work? "Using ray tracing, the Geo Modeler simulates radio transmissions in complex geospatial environments and integrates with multiple internal systems to help automate network decisions and change how the network is configured to improve overall performance," Savoor explained.

Ray tracing is an advanced 3D rendering technology most often associated with video games. Put simply, Geo Modeler uses ray tracing and other technology to create a virtual model of the AT&T mobile network. The new AI system can then forecast coverage gaps and provide solutions.

AT&T provided this example of a Geo Modeler data visualization. Credit: AT&T

Savoor said that the system will still be closely monitored by employees to ensure the AI provides accurate recommendations.

Geo Modeler will also support first responders

After a natural disaster, AT&T supports the efforts of first responders with its FirstNet platform, which the company describes as "the only nationwide, wireless communications platform dedicated to and purpose-built for America’s first responders." Geo Modeler will improve FirstNet network reliability, even in very challenging environments, AT&T says. Likewise, Geo Modeler will be able to optimize the network to make it easier for first responders to deploy drones for surveying or rescue operations.

Geo Modeler will also benefit the AT&T Network Disaster Recovery program, which transports equipment, such as cell towers mounted on trucks, into disaster areas. In addition to adjustments to its radio antennas and cell phone towers, Geo Modeler will be able to provide recommendations to support the NDR program.

Now, in the event of a hurricane, tornado, or wildfire that puts stress on its mobile network, AT&T told Mashable that both customers and first responders are more likely to stay connected, potentially saving lives.

Velin Kounev, the chief scientist on the project, said that Geo Modeler was developed in-house over four years and is already being used in various ways.

“Whenever a tower goes down, we immediately, in 15 minutes, know about the tower going down,” Kounev said. “We know the impact it has, and sometimes the impact of the tower may be hundreds of kilometers away…So on a human level, what that means is that when there's hurricanes, when there's [tornadoes], we can very accurately predict if there's going to be a degraded performance somewhere in our network, and we can also compensate for the degraded performance in real time.”