Nokia’s internally developed ReefShark system-on-chip (SoC) platform is gaining artificial intelligence (AI) capabilities that can help manage 5G radio access network (RAN) operations and further support the burgeoning RAN intelligent controller (RIC) market.

The AI work is under Nokia’s MantaRay umbrella, which includes four specific product focuses: network management, self-organizing networks (SON), energy and RIC. Mark Atkinson, SVP and head of RAN product line management at Nokia, described these as a “suite of AI-based operability and solutions for managing the networks as a whole, like radio access networks.”

MantaRay Network Management is the “future platform” for Nokia’s extensive purpose-built RAN, cloud RAN (cRAN) and core network management products designed to “scale from a single server for small enterprises to large-scale implementations.” This would be from small-scale private network deployments up to nationwide 5G networks.

“We're bringing more AI-based capabilities into MantaRay Network Management over time with a view that networks are becoming so complicated that it's impossible for human engineers to manage them in the most efficient way,” Atkinson said. “We used to do a lot of things based on automation, though the idea of automation is that you're automating human tasks, but they're still task based, whereas the AI and machine learning aspect is that the system learns over time how to best optimize performance itself.”

Beaming with AI and automation

MantaRay SON uses AI and cognitive capabilities to help operators automate the setup and management of their network. This includes the use of intelligent beamforming and beam settings.

“Every cell site needs to know what's going on in the surrounding cell sites so that it knows when to hand over a connection from one cell site to another,” Atkinson said. “Having that manually created as databases is a massive workload as things change in the network all the time. The more you can automate and then adapt accordingly helps the overall efficiency of the system. It's about self-configuration, self-optimization, self-healing, so if one cell goes down, another cell picks up the traffic. We're evolving that with more AI-based capabilities.”

This also feeds into the energy component that uses AI to dynamically manage base station power requirements. This is a step beyond traditional methods that typically have set power cycle management, like reducing power during overnight periods when base station load is typically lower.

“Ideally, what you would want to be doing is that on a completely dynamic and learning bases, where each base station can fire up cells depending on traffic needs,” Atkinson said. “If only one device is on a particular cell you don't need to have five radios pointing at that device with carrier aggregation and all of those things. One will be enough and so it's about how we best utilize the capabilities that we have in the radio software and optimize accordingly.”

Atkinson also noted that MantaRay can handle this level of management across different vendor equipment if that equipment has the appropriate software to allow for that control. “Tier-one suppliers like ourselves and Ericsson, for instance, will have a pool of those capabilities and we're able to manage both networks,” he said.

The final component is the RIC, which is an area of intense development. Nokia’s MantaRay work is on the near real-time (xApps) space that Atkinson said provides millisecond-like control “in a cluster of base stations or sites with specific profiles,” however, he also noted that this work is still in its infancy.

“This is an investment in the future,” Atkinson said. “We're still learning here and we've had trial projects with customers and got some good feedback. Now we need to figure out how to scale it how to make it into a commercial valuable model for the customers and for us.”

Nokia AI for ReefShark and RAN

Nokia’s RAN-focused AI work is built on ReefShark architecture that dates back to 2018. Those efforts gained a boost in 2020 when Nokia tapped Intel and Marvell to help diversify development, and commercial products hitting the market in 2021.

U.K.-based telecom giant Vodafone last year added Nokia’s ReefShark platform to power commercial-off-the-shelf (COTS) servers supporting the carrier’s open RAN infrastructure.

Dell’Oro Group recently lauded Nokia for outgrowing its RAN rivals during the first quarter of the year.

Stefan Pongratz, VP at the analyst firm, noted Nokia “recorded the highest growth rate among the top five suppliers,” with its RAN revenue share outside of China now “trending upward over the past five quarters.”

Despite that growth, Nokia remains behind Huawei and Ericsson globally and behind Ericsson among RAN suppliers outside of China. Samsung earned the No. 5 position globally behind ZTE, but jumped a spot to No. 4 when taking China sales out of the equation.

Nokia’s RAN market surge followed a robust quarter of news from the vendor.

Nokia posted a 10% year-over-year increase in total net sales for the first quarter and a more substantial 32% increase in net profits. That growth was just behind what Ericsson managed to post, but Nokia’s net income far surpassed the 46% year-over-year drop reported by its rival.

Nokia President and CEO Pekka Lundmark touted strong growth in 5G equipment sales across some markets, specifically pointing to India and “good traction in Europe.” Lundmark added that Nokia had around 30% market share across Europe at the end of the quarter, but was gaining half of all new contracts tied to operators ripping out equipment from China-based vendors like Huawei and ZTE.

Despite its momentum, Nokia’s management provided mixed messages on how the rest of the year is going to play out for the telecommunications infrastructure and software vendor.

Lundmark said the vendor was “starting to see signs of the economic environment impacting customer spend.” But, he added that customers still had a need to maintain their investments in 5G and fiber, with questions remaining over the timing of any spending impact.

“It’s clear that there is some economic uncertainty impacting customer spending plans,” Lundmark said. “In that regard, it is worth noting that if we look globally, excluding China, only about 20% of sites … are currently active for mid-band 5G, which should help illustrate how much investment still needs to be made. And even if we look at some markets like North America, which had invested earlier, mid-band site penetration is still only around 50%.”