Your browser is not supported. Please update it.

13 September 2021

Open RAN Summit: MNOs point to rapid progress but many remaining challenges

Any new and disruptive technology will have teething problems, especially one in which many operators and vendors are trying to assert their influence, and one that carries the weight of political expectations also. Open RAN is such a platform, and continues to attract its share of scepticism – some commercially motivated, some very valid – while also building up its support base.

One complication is the presence of Chinese intellectual property and member participation in the O-RAN framework, the most prominent set of specifications underpinning the push towards an open disaggregated RAN. This works against the US-inspired narrative that Open RAN will be a way to exclude Chinese technology (and alleged spyware risks) from 5G networks and build local ecosystems. It was highlighted when Nokia suspended its O-RAN activities in order to address possible conflicts with US rules.

Such difficulties are likely to be addressed over time, or the impetus may shift to other open RAN efforts such as those housed by Small Cell Forum and Open Networking Foundation (the latter O-RAN based, but capable of becoming more standalone). Dennis Hoffman, general manager of Dell Technologies’ telecom division, made it clear, in a recent presentation, that the bid for open networks must not be confined to one organization, whether a single vendor or a cooperation.

“The bigger picture isn’t so much about any one company, nor is it necessarily about any one alliance or standards effort, because the mega-trend is marching on, and I don’t think it’s going to slow,” he said. “If anything, we’ll see a typical ‘S’ curve where it will bounce along at a certain rate and hit an inflection point and then it will accelerate pretty rapidly.”

He said in an interview with LightReading: “It’s not surprising that some of today’s networking equipment vendors are not fans of O-RAN, the way it is described, because it really pressures their business model … But as we’re seeing it, across the entire industry, there’s tremendous innovation going into every level of opening up the network.”

Dell’s immediate interest in Open RAN is its potential to accelerate the virtualization of RANs – the most challenging aspect of the broader move to software-based networks whose functions can run on generic cloud hardware. That can provide a new market for Dell servers, and the company already has commercial vRAN deals with operators like Dish. But over time, it also wants to build a top-to-bottom Open RAN business in which it sells software and systems integration services.

Dell takes a more positive view than Nokia, it seems, about the significance of the O-RAN Alliance having three Chinese members that are on the USA’s entity list. Hoffmann said this had not slowed Dell’s involvement in O-RAN activities, even though as a US firm, it would be bound by Department of Commerce rules that forbid trading with an entity listed company without a special licence.

Operators seem less concerned about the politics of trade sanctions and more about the challenges of making an open, virtualized architecture fully robust and capable of supporting large-scale 5G macro networks in future.

Though some large non-greenfield operators, such as AT&T and Vodafone, plan commercial Open RAN deployments in the next 1-2 years, these are initially focused on relatively undemanding rural or small cell networks while most telcos see 2025 or later as the timescale for fully loaded macro 5G.

Deutsche Telekom recently provided its view on the Open RAN progress report, saying the pace of development was encouraging, but the platform in its current form is too complex. Essential capabilities in programmability and automation are 2-3 years away, the operator said.

Franz Seiser, VP of access disaggregation at DT, spoke at the recent Fierce Wireless Open RAN Summit, and reiterated the firm’s “very strong belief” in Open RAN. “As we all know, it is not fully there yet; there is still a way to go,” he said, citing significant technical and cost issues related to the complexity of the platform and the need for integration services.

However, while it would not, in its current state, deliver total cost of ownership (TCO) savings, he said: “We see how the price is developing in the right direction, and the really good news is it’s developing even faster than we hoped or anticipated. Due to the openness, having more companies, and innovating at a much higher pace on the semiconductor and silicon level, we see things accelerating, and of course, that helps to work on the TCO.”

Before Open RAN will be truly transformational, it needs to be fully “software‑centric, flexible and dynamic” and to “translate the radio into a programmable platform”, Seiser added. In this, the RAN Intelligent Controller (RIC) will be key, but he thinks full programmability is 2-3 years away.

“First of all, we need to get this thing implemented … it has to deliver, it has to compete on legacy. If this is not achievable, then it will be tricky”, he said. Integration costs and efforts are currently too high, as very few operators will want to emulate Rakuten’s expensive deployment, supported by several hundred engineers. “We see that complexity on our side, while at the same time we have cost pressure,” said Seiser. “I cannot say I need a team of 200 people to do the integration; that’s impossible.”

Automation will be a “must-have”. He also acknowledged that, in DT’s experience, it was not yet true that the RAN software could run smoothly on any hardware, and pointed to difficulties when there are hardware or driver updates, or new software.

Another essential milestone will be the integration of Massive MIMO with Open RAN. Despite some progress in tests and trials, this is “really complex and requires a lot of fine tuning”, Seiser said. DT has been running Open RAN Massive MIMO units in its ‘O-RAN Town’ in Neubrandenburg, Germany, where it is trialling the technology over a projected 25 4G and 5G sites.

Such doubts are often seized upon by sceptical or hostile stakeholders as proof that Open RAN will fail, while in fact they represent the inevitable early challenges that even enthusiasts like DT will meet when trialling a very new architecture.

While Chinese players, notably China Mobile and its ecosystem, have been very significant contributors to Open RAN, and triallists of the technologies, Huawei is not getting on board. Its CTO, Paul Scanlan, made one of the clearest statements to date that the giant supplier will not support Open RAN, on a media call last week. He said Open RAN was not standardized, could not be easily integrated with existing network infrastructure, and is not ready for large-scale 5G.

He echoed the concerns of some other operators and vendors, that Open RAN in its current form will get its timing wrong in macro networks, though likely to have a strong base in enterprise and small cell roll-outs. “It’s not that it’s not going to happen, and I believe it will in different guises but I’m not sure whether … from a commercial perspective, is it too late practically?” Scanlan said.

The O-RAN Alliance is not a standards body but “just a bunch of friends”, he said, claiming that to succeed, Open RAN must be pushed into the 3GPP to expose it to representatives from about 200 countries, increasing innovation and avoiding fragmentation.

Huawei is also negative about vRAN architectures in general, whether open and multivendor or not. Scanlan believes virtualization will be important in some parts of the network such as the 5G core and the mobile edge, but questions the performance and security issues associated with the vRAN.

“For virtualized RAN, what do you want to do with virtualization, what’s the target objective?” he asked. He questioned the quality of service and performance that could be supported by commodity hardware. Huawei trialled commoditized chips in 5G equipment back in 2017, he said, but “the problem was that the jitter at the substrate level was too high. It would not achieve the targets that we wanted in terms of latency, so we had to develop the chip ourselves”.

But the US giants, Verizon and AT&T, both defended Open RAN, at least in limited scenarios, at the FierceWireless event. Raj Savoor, AT&T’s VP of network analytics and automation, said the firm would start indoor deployments of Open RAN in about a year, since trials were “giving us confidence” in the technology. He said the software that would “unlock the capabilities” of Open RAN is still in development, especially the RIC.

Adam Koeppe, SVP of technology planning at Verizon, said his company is “focused on commercializing an operationally sound Open RAN architecture”. He said the main barriers were resistance from some existing suppliers, the R&D burden that was placed on small suppliers, and the need for standardization of management interfaces and a structured approach to ongoing enhancements and updates.

He also warned: “There will be increasing complexity and integration costs associated with integrating different components at scale.” However, Verizon expects to deploy some Open RAN at the network edge in late 2021.