The third annual Telecom Infra Project (TIP) FYUZ event rekindled open RAN at a time when some were beginning to question if the approach had lost momentum, with organisers firmly consigning the flashiness of the first edition to the history books in favour of a tighter, more focused agenda sporting top-tier executives and experts.

FYUZ24 found a new home in the Republic of Ireland’s capital city Dublin, a move placing the event closer to the heart of a major urban centre than its original location on the outskirts of Madrid and enabling local speakers to attest to the benefits of digitalisation enabled by more ready access to connectivity.

But it was perhaps the tangible, albeit apparently quiet, progress being made by the open RAN community as a whole which gave the event more bite than the culinary-themed original in 2022 or the mastication of key challenges in 2023.

A prime example of the progress of open RAN as a whole came in a panel featuring speakers from AWS, Boost Mobile and Accenture, which addressed the nuts and bolts of a real-world deployment and the lessons learned.

Hands on
AWS was Boost Mobile-owner Dish Networks’ provider of choice when it began building its cloud-native 5G network, so Kaniz Mahdi, director technology at AWS Industries (pictured, far right), and Satish Sharma, EVP network development with Boost Mobile (pictured, second from left), offered insight into their progress and how open RAN is working out in a real-world deployment.

Mahdi explained AWS now operates 19 sites with a centralised RAN serving distributed users: “it’s working, so it is the reality, the future is now and we are now building the next chapter”.

The executive noted this is a far cry from prior TIP and other events on the subject, where “mostly you see demonstrations, experiments”, progress which is helping to define the next steps.

She identified “speed, scale and security” as challenges for which “AI and data have been very helpful” in addressing.

Mahdi conceded AWS’ work with Dish Network had not been smooth: “we did it in a rather scrappy way”, she said, noting the companies even identified some gaps in network observation capabilities on the day of the presentation.

She explained hindsight is always 20:20, stating if the companies were to commence the task today, they would focus more on the ecosystem and “ironing out the kinks” before placing anything in the cloud.

The expert noted working with standards bodies including the O-RAN Alliance is, of course, a key element, but added preparing for the next “iteration of RAN evolution” should ideally involve creating “communities like this where we are having a multi-sector discussion” and implementation, so “the cloud providers and the workloads that go on the cloud” are grouped together to enable companies to activate services with a flick of a switch without teething troubles.

“I would say the most important thing would be crowdsource innovation and bring multiple sectors together in the same community.”

Intrinsic
Sharma asserted Dish Networks’ progress in its greenfield 5G deployment was better today than it would have been using more traditional methods, branding it “one of the largest deployments” of open RAN in the world.

The network has 250 million POPs, a feat only achievable because it lacks “any legacy” equipment. Sharma said the progress is “mind blowing”.

It faces several challenges in delivering on the potential of open RAN: Sharma pointed to operating costs as a primary consideration, along with efficiency in terms of how best to employ the vast volume of data generated by its open interface.

“There are multiple layers of insights”, Sharma said, explaining this feeds into network performance, but also the customer experience provided.

Of course operators are keen to work out how to generate revenue from open RAN infrastructure: “all that talk about technology and, as an operations person I can speak about operations, but unless we get the revenue”, a growing, satisfied customer base and an ability to “demonstrate different use cases to enterprises” in a large-scale way, “until that, it’s just a story”.

AI comes into play here, in terms of how the operator connects elements including the RIC, rAPPs and xAPPs “and everything else that is coming together”, to then take it to the next level and enable monetisation “in a big way”.

The next three-to-five-years will see the company focus on ecosystem development, ensuring technology and products truly become plug and play.

“For that to happen, there’s a level of work that the stakeholders’ group, the OEMs, the people who provide the product and technology really need to spend time making…mature enough” for operators to run out of the box.

Boost Mobile is keen to see a “larger ecosystem” of partners take “software to a level of maturity and resiliency that, once we put it in production, really works”.

AI
Mahdi agreed AI is pivotal to best employing network data in distributed networks which may span several nations and regions of the world.

The technology can be used to deliver a “control plane” which goes beyond the RAN by incorporating the “intelligence that is driving the network as well as the user” to provide a quality service.

One factor which could come into play is the investment cycle anticipated when 6G begins to become a reality, though Mahdi argued there is still time for open RAN to be prioritised before the next-generation technology begins to take shape.

Jefferson Wang, CSO, cloud first at Accenture (pictured, second from right) argued AI is emerging as an important element in terms of extracting maximum value from open RAN deployments, opening the door to smoothing repetitive network processes and a shift “from reactive monitoring to proactive and predictive maintenance”.

There is also a role for AI in accelerating development and deployment of xAPPs, Wang said, pointing to the potential to cut the time taken to test RIC software.

Wang said fresh challenges are coming to the fore, with edge processing of open RAN posing questions around which elements would be handled on site compared with being passed to the centralised cloud.

Security remains a key consideration for any disaggregated network play and, again Wang sees a role for AI to address questions around network segmentation and assess the “blast radius” before committing to any changes.

Wang highlighted some more concrete problems involving operating models, because the increasing number of players involved, “whether it’s third parties” or others, is resulting in “incredibly complex” structures.

Ironically, the technology advances mean the skills of staff remain as important to a successful open RAN deployment as the infrastructure itself, Wang added, explaining just hiring new people is not necessarily the best way to accrue the skills required.

Accenture views this more as a “kind of talent creation, where you’re going to actually have to take a network engineer and train them more on software versus going the other way” by trying to teach a software specialist how networks operate.

Wang believes creating a testing environment focused on interoperability “is the first big important piece that we hope to accelerate” in future, though also highlighted the compute and storage power required to fully employ AI, along with arguing the whole movement cannot be solely centred around the cost of ownership.

The panel may have highlighted plenty of current and emerging challenges remain around adoption of open RAN and broader network disaggregation moves, but the vibrancy of FYUZ24 also highlighted the progress to date and how the community itself seems to truly be finding its feet.