Is the end of weak signals in sight? Nokia and O2 have deployed artificial intelligence directly into a live 5G network

You know the feeling—you’re standing at a crowded festival, in a stadium, or at a train station; your screen shows a full 5G signal, but you still can’t send a single text message. The network is simply overloaded. However, telecommunications giants Nokia and Virgin Media O2 have discovered that artificial intelligence can solve this problem in real time. They recently announced the successful testing of the groundbreaking AI-RAN technology in live operation at one of Europe’s busiest locations. The results suggest that the era of unstable internet and “dead zones” may soon be over.

While operators have traditionally relied on building new physical antennas or raw computing power to improve mobile networks, the new era belongs to software and machine learning. A technology called AI-RAN (Artificial Intelligence in Radio Access Network) is bringing a massive change to the world of telecommunications.

Instead of antennas merely statically transmitting and receiving signals from users and only reacting when the network goes down, artificial intelligence enables the radio network itself to “think” and anticipate events before they even occur.

Real-world test at a train station: 20% higher efficiency

To prove that AI-RAN is more than just a lab experiment, the companies deployed it at an O2 “Giga Site” transmitter right in London, specifically at the Paddington transportation hub. This is one of the most challenging environments for any mobile network—passengers arrive in huge waves, trains are constantly emptying, and thousands of smartphones begin competing for data capacity in a single second.

The results of a recent test exceeded expectations. According to official measurements, Nokia’s artificial intelligence was able to increase the network’s spectral efficiency by up to 20%.

In practice, this means that without the operator having to add new radio frequencies or install additional antennas, the existing transmitter was able to handle a fifth more data traffic. It’s equivalent to a situation where, thanks to smarter traffic management, you could squeeze 20% more cars onto a three-lane highway without causing a traffic jam.

How exactly does AI correct the signal?

A traditional radio network operates according to relatively fixed rules. The transmitter uses static algorithms and cannot respond flexibly to emerging chaos.

Artificial intelligence in the AI-RAN system eliminates this problem. It collects millions of data points in real time about how users are moving and what apps they’re currently using. Based on this data, it can:

• Predict congestion: The software constantly learns from changing conditions (such as the arrival of a train). Artificial intelligence redirects capacity and frequencies to the hotspot even before people start downloading maps or uploading videos en masse.
• Automate without human intervention: Radio resource management occurs dynamically in the background, completely eliminating the need for manual intervention by the operator’s network engineers.
• Stabilizing response time (latency): Even during extreme traffic spikes in London, the network maintained stable accessibility and reliability—a benefit particularly appreciated by mobile gamers or those making video calls, where the system creates a “smoothed” data tunnel.

Benefit for users: Better internet without investing in a new phone

The most important aspect of this project for ordinary consumers is that they don’t need to buy any new hardware to transition to AI-RAN. You don’t need a new smartphone labeled “AI-Ready.” It’s the invisible infrastructure above you that’s being improved.

“Demand for mobile data is constantly growing, so we need smarter ways to get the most out of the spectrum and infrastructure we already have,” said Jeanie York, Chief Technology Officer (CTO) of Virgin Media O2, as part of the project. According to her, the project has proven capable of serving customers more efficiently without any increase in power consumption.

The deployment of AI-RAN in the UK network is part of a massive £700 million investment plan, but Nokia is also gradually rolling out this technology to other operators. Although widespread global deployment across all European (and eventually Slovak) cities will still take some time, this is clear evidence that the true benefits of artificial intelligence are no longer limited to our screens but will ensure the smoother operation of the entire digital society.