T-Mobile CTO on injecting AI into network resiliency

5 Min Read

A satellite-equipped mobile cell site truck was deployed to the Walton County Emergency Operations Center in Florida to assist with Hurricane Isaias recovery efforts.T-Mobile

For organizations that provide essential services and operate in remote areas, reestablishing network connectivity is critical when disaster strikes. A hurricane such as Isaias, which made landfall in Florida last week, can damage the electrical grid and cell sites, resulting in major disruptions to business and network operations for CIOs and communications challenges for first responders.

Organizations such as hospitals, government agencies and financial institutions require network restoration as quickly as possible, and rely on the latest network technology to ensure efficient recovery. Mobile network operators (MNOs) are now adding AI — with technologies such as network slicing and satellite backups — to ensure faster disaster response times and business continuity for CIOs leading critical teams.

CIOs overseeing essential services are primarily concerned with what Omdia analyst Ildefonso De La Cruz Morales refers to as The Big C's: Coverage, capacity and congestion avoidance. They want assurance that in the event of a disaster, they will still have access to a high-performing network that isn't slowed down by traffic congestion.

Related:CISOs shouldn't sleep on banning smart glasses

"One of the major concerns is how can we guarantee that they get the resources they need when they need it," he said.

Network resiliency has come a long way in recent years, as MNOs such as T-Mobile build out their 5G networks, provide dedicated network slices for first responders, and deploy physical assets, including cell on wheels (COWS) vehicles and cell on light trucks (COLTs) to establish temporary network connectivity. AI is taking network restoration one step further by minimizing time to action and ensuring emergency response decisions are based on as much information as possible, according to De La Cruz Morales.

In an emergency, T-Mobile deploys physical assets, including COWs and COLTs, to provide cellular backhaul and temporary network coverage, according to John Saw, CTO at T-Mobile.

Many of T-Mobile's cell sites have new hybrid generators for longer runtimes, and the company taps satellite connectivity for backup network coverage, according to Saw. Earlier this year, T-Mobile launched "SuperBroadband" for businesses, a managed hybrid internet service that delivers fixed wireless 5G as a primary connection and satellite broadband from Starlink as a backup.

"We're bringing in a layer of AI on top of all these assets, so that we can make better use of them, deploy them more efficiently, and also deploy them more adaptively where the help is needed the most," Saw said.

Related:The CIO's maintenance backlog is now a business risk

One of T-Mobile's SuperBroadband customers is Aramark Destinations and Aramark Nexus, divisions within Aramark that provide hospitality services to vacationers and workforce services, respectively. Dimple Jethani, CIO of Aramark Destinations and Aramark Nexus, explained that having network redundancy — through fiber, cellular and satellite connectivity — is critical to providing a seamless vacation or work experience.

"Having this managed service, a singular way to monitor it and a single way of operating has created a lot of ease and opportunity for our business," she said.

Many of the properties Aramark provides hospitality and workforce services to are remote and spread out, but the company's goal is to provide a "like-home" experience in terms of connectivity, she said. Aramark Nexus provides workforce services for data center construction workers operating in remote areas, and providing them with the level of network access they're accustomed to at home — like video streaming services — is beneficial for worker retention, Jethani explained.

Automating network resiliency with AI

T-Mobile also has a dedicated network that prioritizes connectivity for first responders — "T-Priority" is a network slice supported by T-Mobile's 5G standalone core network. The other two major U.S. wireless carriers have their own existing services that prioritize connectivity for first responders — Verizon Frontline and AT&T's FirstNet.

Related:Cyber risk falls flat without business translation

T-Mobile recently added AI capabilities to make network adjustments faster — such as automating antenna tilts and power-level changes. AI can anticipate traffic demand and automate the optimization of network performance, according to Saw. The new AI capabilities also provide intent-based AI automation to identify and deploy network adjustments based on desired outcomes.

"We can basically adapt the network during natural disasters when every minute counts," he said. " When we recover the [cell] sites or the site comes back after power is recovered, then it will automatically adapt back to its original optimized status on a clear day."

During winter storm Fern in January, which spanned from Northern Mexico to Canada, T-Mobile used intent-based AI automation to make 30,000 antenna adjustments. Making that many real-time antenna adjustments wouldn't be possible if they were performed by human engineers alone, he explained.

T-Mobile is also using AI to automate network adjustments during periods of high traffic demand, ramping up capacity for sporting events like the recent World Cup. AI is used to forecast and make network adjustments on cell sites based on where fans gather to "make sure that we have the highest capacity possible, and then AI adjusts it back to its normal settings once the crowd disperses," Saw said.

"That's one good example that's fairly new, basically using AI to anticipate intent and what our customers are intending to do, and then taking adaptive steps to get us there without engineers getting into the loop and slowing things down," he added.

AI can be used to process incoming data "into actionable steps in real time" without overwhelming first responders and creating decision paralysis, De La Cruz Morales said. However, it's still early days for AI solutions within the disaster preparedness sector, he explained. Whether the compute process for AI is performed on devices, at emergency response hubs, or via satellite, for example, is still being worked out.

"Where does all the processing happen with AI? That's still not decided," he said.

Is your organization using AI for network resiliency? Let us know at [email protected].

About the Author