Nokia is not holding its traditional Sunday afternoon media and analyst conference prior to MWC this year. After going large on a major strategic and brand launch (T-shirts and all) last year, this year it has discontinued the pre-MWC custom – although it has kept one tradition alive, launching out press releases like a hyperactive (branded) T-shirt cannon operator.
It has also produced a paper on its O-RAN roadmap that, at times, carries the passive aggressive energy of a wronged partner.
Wronged? Well, Nokia may insist that its relationship with AT&T is still strong (it sells lots of optical and IP infrastructure to AT&T) and it may have wrung out of A&T the concession that Nokia didn’t lose out in a technical beauty contest, but the impact of that decision clearly still smarts – especially with those on the radio side – and Nokia has decided it needs to tell its truth to the market.
In a LinkedIn post quoting a report on RAN market shares from analyst company Dell ‘Oro, Nokia’s head of RAN Network business unit Mark Atkinson said, “There continues to be a lot of opportunities for growth when you have a strong technology position, services capabilities, industrial scale and customers who make well informed decisions.” Ouch, what could he mean? Well one current lack of information, as Nokia sees it, may lie with those that are currently giving Ericsson credit for being an enabler of O-RAN rollouts – there have been recent endorsements not just from AT&T but also Vodafone and Telefonica.
The new position paper from Nokia, released right before MWC, makes its own commitment to an O-RAN roadmap clear and affords it the opportunity to take a couple of thinly veiled swipes at what Nokia clearly believes is Ericsson’s lack of O-RAN readiness.
As it listed off evidence of its own commitment to an O-RAN roadmap (more on that later), the Nokia paper said that some industry suppliers claim O-RAN leadership on paper, but have yet to prove third party RU-DU integration, nor do they have an actual multi-supplier O-RAN deployment in a live network.
Nokia for its part has public integrations with Fujitsu, Mavenir and Samsung RUs, and says it has completed two further integrations that are not public for now. It is also integrated with third party RUs in commercial networks with DT and NTT DoCoMo.
Ericsson has no such public references, save the fact that AT&T has confirmed that Fujitsu RUs will be a part of its network refresh with Ericsson. Ericsson told media at its pre-MWC event that that it doesn’t like to crow about each individual integration it does, and later added to TMN that it has in fact been doing more RU-DU integration than may be obvious. It added that it would have evidence of that on its booth this week. One to look out for.
Show your EQ
Aside from asking operators to look for true evidence of RU-DU integration, Nokia goes deeper into the interoperability weeds, pointing out that operators should also task vendors with ensuring that O-DUs support both flavours of the recently specified ULPI (Uplink Performance Improvement) fronthaul specs.
ULPI is the interface specification that has been designed to enable more efficient operation of massive MIMO across O-RAN fronthaul (between the DU and RU). It is the spec that Ericsson has made a great play of leading and pushing through O-RAN specs, and has allowed it to say it has a million radios already deployed that are hardware ready to support Open RAN interfaces.
And yet Nokia is indicating that it was equally engaged in and supportive of ULPI. Further, it points out that not all suppliers – it doesn’t name Ericsson but who else could it mean, really – are fully committed to implementing both deployment variants of ULPI.
Yes indeed, if you weren’t aware, there are two deployment options for the ULPI fronthaul interface – Class A and Class B. Both Class A and Class B place Channel Estimation and Interference Rejection Combining (IRC) at the RU. Class A also places the equaliser at the RU, and bypasses the equaliser at the DU. Class B keeps the Equaliser only at the DU, and doesn’t implement it at the RU.
Following disagreement through 2023 between two camps within the O-RAN Alliance as to which option was preferable, it was agreed as a compromise that anyone who deploys Class A should also support Class B, even if the Equaliser was never activated at the DU. Nokia counsels operators to make sure that vendors support both options. It says this is an important contributor to openness because it means any class of RU can be connected to the O-RAN network, rather than solely the more advanced Class A RUs.
In other words, were a DU vendor to go to market supporting only Class A (ie no support for the equaliser in the DU) then it would not be fully O-RAN compliant, and would be depriving operators of RU vendor options.
Now, Ericsson has been clear it will support Class A – it said as much to TMN last week in this piece. Nokia is making a public commitment to support both Classes, and it wants operators to know and understand the difference. And, presumably, to pressure Ericsson to also support both classes.
Cloud RAN migration path
Nokia also diagnoses a lack of a clear commitment to software feature and performance parity between Cloud RAN and purpose-built RAN. Note that Cloud RAN (vertical disaggregation) doesn’t have to equal Open RAN (horizontal), the two are separate development paths.
The paper says that some vendor plans have “not been backed by a clear commitment to software feature and performance parity between Cloud RAN and purpose-built RAN. It remains to be seen if or when this will be observable with these suppliers.”
Nokia itself lays out how it will migrate to a Cloud RAN architecture that also supports Open RAN interfaces.
It says its first step is to connect O-RUs to its dedicated (ie not Cloud RAN) baseband. This it has already done. The next phase is to deploy its Cloud RAN architecture with eCPRI (7.2x) support to its own RUs. It said it will have this capability in 2024. A third phase will see its Cloud RAN be able to connect over Open Fronthaul to third part vendor RUs – this happens in 2025. At this stage, Nokia will also offer ULPI connectivity to its own mMIMO antennas (for Massive MIMO use cases, the plans for Nokia AirScale Habrok radios include O-RAN eCPRI 7-2x ULPI uplink and Cat-B downlink support.)
A final step, to happen from 2025 onwards, will see Nokia’s Cloud RAN be able to connect to third party mMIMO units. Note that this means that Nokia is saying it will be 2025 and after before its Cloud RAN variant will ahve support for third party mMIMO.
The vendor says this approach will keep feature parity between its Cloud RAN and dedicated RAN options at all stages of development and deployment. It is also a pragmatic approach, Nokia says, that recognises that operators will have hybrid cloud and non-cloud networks, and hybrid open and non-open networks, while wanting feature and performance parity across all.
“Some CSPs will deploy Open Fronthaul without Cloud RAN whereas other CSPs will deploy Cloud RAN without Open Fronthaul. Most networks will eventually become hybrid RAN networks with the coexistence of Cloud RAN and purpose-built RAN using various O-RU options in different geographies and/or frequency layers.”
