AI EDUCATION: What Is AI-RAN?

Have you ever notice how many abbreviations and acronyms are tossed out by technologists industry news writers with nary an explanation?  

I guess it’s a good thing that I love being stumped and confused, especially when it comes to technology, as it sends me down a rabbit hole of discover. Which, I guess, is why this week on AI Education we’re going to discuss the promise of AI-RAN, or artificial intelligence radio access networks. 

Of course, to get things started, we’ll probably want to explain in brief what a radio access network is to make sure your eyes don’t glaze over midway through our column. A radio access network is the computing and communications infrastructure that resides between mobile devices and a core network. In particular, a radio access network encompasses all of the antennae that our phones and tablets and laptops connect to and the convoluted system that helps our devices remain connected in a manner that appears to us as completely seamless as we move. 

Still with me? Good. An artificial intelligence radio access network, then, is just integrating artificial intelligence capabilities within those radio access networks. As telecom industry news provider Fierce Network puts it, AI-RAN is not a single technology but the integration of AI and machine learning with digital telecommunications technology. It turns out there are some significant benefits to mixing the two. But first… 

How Did We Get Here? 

So it’s been about a month since the report came out, but we stumbled upon a discussion of AI-RAN in another of those predictions for the year to come reports about 2025, this one from ABI Research. Unlike most beginning of the year predictions, ABI was trying to guess what wouldn’t happen this year. 

Among the things ABI doesn’t expect to happen are: smart glasses (still) won’t see widespread adoption, AI-driven humanoid robots won’t take over the world, and AI-RAN will not be widely deployed. 

From ABI’s analysts: “Despite rapid progress, Artificial Intelligence (AI) technology for the RAN ecosystem (AI-RAN) remains in early development, with commercial deployments not expected before 2026. SoftBank recently completed a trial and plans deployment by 2026, with broader adoption likely by 2027, given Japan’s history as an early tech adopter. In 2025, trials and pilot projects will advance the technology, but large-scale deployment will take longer to materialize.” 

Earlier in 2024, technology and telecommunications leaders formed the AI-RAN alliance to advance work on artificial intelligence in radio access networks. Founding members included Amazon, Microsoft, Ericsson, Nokia, Nvidia, SoftBank and T-Mobile. The mission, at launch, is to “enhance mobile network efficiency, reduce power consumption, and retrofit existing infrastructure, setting the stage for unlocking new economic opportunities for telecommunications companies with AI, facilitated by 5G and 6G.” 

How Do RANs Work? 

You’ve probably heard a lot about radio access networks over the past 15 years while rarely hearing the term directly mentioned, if at all. Every time you hear the terms 3G, 4G, 5G—and soon-to-be 6G—mentioned in the context of wireless communications, what you’re hearing discussed are different generations of the RAN. 

RANs keep us connected. As 5G communications have been deployed, RANs keep us connected with near ubiquitous coverage and at very high speeds. RANs work by receiving signals from our device, translating the data from that signal into a radio format, processing that data so that it can be communicated with the central network, then sending the information to the central network where more sophisticated functions can be rapidly performed—and then going through the process in reverse to send data back to our device. 

A RAN is able to perform work for many devices simultaneously on a very large scale, however, with the advent of 5G, RANs are being asked to perform continuous, simultaneous work for an ever-growing universe of devices. 

Why Is AI-RAN Important? 

So it’s not going to happen in 2025 (if ABI Research is to be believed, anyway). AI-RAN is still important to understand because it will solve huge inefficiencies and vulnerabilities in communications infrastructure, according to the AI-RAN Alliance. As demands on wireless networks become more voluminous and complex, conventional infrastructure, even 5G infrastructure, becomes a bottleneck—systems can struggle to link devices and networks continuously. Think of an air traffic controller with too many flights on their screen. Furthermore, trying to move so much data through conventional infrastructure becomes increasingly energy inefficient. 

AI-RAN, with machine learning and decisioning abilities, can centralize all of the functions of an RAN (including networking, computing and storage) into one infrastructure. AI-RAN is capable of allocating computing resources to optimize efficiency, and handling the mushrooming number of devices with wireless connectivity. This deployment is sometimes deemed “AI for RAN.” 

Better radio access networks, and more connectivity between our devices, also makes possible the increased utilization of distributed computing, virtual supercomputers built not in a centralized location, but across a cloud, cooperatively, by users who may be geographically distant from each other. 

As with any AI deployment, AI-RAN will also make available to technology providers and telecommunications companies a trove of data that can be harnessed to offer new services and insights, and potentially monetized into new revenue streams.  

AI-RAN As a Foundation for Telecom’s Future 

According to Nokia, “6G will not work without AI.” The next leap forward in connectivity, 6G, will be an AI-native infrastructure. AI-RAN is also the key to deploying AI both centrally, on a network accessed by our mobile devices, as well as along the edge, on our devices themselves, and enabling those different levels of AI to communicate and collaborate. Some industry writers and observers deem this “AI on RAN.” 

The long-discussed Internet of Things (IoT) will be dependent on AI-RAN. AI-RAN, when implemented, will enable the safer operation of fleets of autonomous vehicles and sophisticated drone networks.