Which way is the pressure switching on Open RAN?

Levels of virtual, levels of open.

Two of the major vendors this week announced further steps in vRAN and Open RAN commercialisation.

Samsung joined Nokia in announcing a vDU (virtual Distributed Unit), the part of a virtualised split baseband that can sit further out towards or in the edge of a mobile network and houses the lower layers of the BBU. (Its heftier counterpart to which it is connected over a “midhaul” interface is known as a vCU – Central Unit. The Central Unit houses more of the upper layer functionality).

Samsung’s vDU will be available later this quarter, and there will be some US field trials after that.  The company said its vDU software can be deployed on COTS x86 hardware.

A press release also also added, “Samsung’s vRAN operates with or without hardware accelerators depending on factors such as total bandwidth.”

That’s slightly more equivocal than a supporting video from the company which said, “Samsung has been developing a fully software-based RAN, with no accelerators or other specialised compute platforms necessary.”

So, can we provide any further clarification on this point? Alok Shah, Vice President of Networks Strategy, Business Development, & Marketing, said to TMN, “The use of hardware accelerators depends on factors such as the channel bandwidth being processed. In some cases, all of the L1 processing will be performed on the x86 platform with no additional acceleration. The default for acceleration will be FPGA, but Samsung is evaluating other processor types as well. Note also that Samsung continues to offer architectures based on our 5G modem ASIC (S9100) for very wide channel bandwidths that may not be cost-effective to implement in software.”

So, that’s not quite as open-and-shut as the promo video (see subtitling in screengrab below), and it is also more detailed than the press release. Shah’s response makes it clear that the “no extra hardware required” only applies in some cases. Once we get into higher channel bandwidths then we leave that behind. And in some instances, you are looking at the dedicated processor solution.

So that’s the vRAN – not quite as native on x86 as some of the publicity around the release stated. What of Open RAN, words that were actually missing from the Samsung press release, but which are very much part of Samsung’s overall messaging?

Here’s Shah again, “Our fully virtualised RAN solution supports our growing portfolio of O-RAN compliant radio solutions.”

How to parse that? Well, first off, clearly you’d expect Samsung’s vBBU to support its own radio units. Given the chart above, you’d also expect it to support radio units from other vendors, over that open Fronthaul interface, and of course Samsung is going to want to interface with vendors of RRUs that support bands that Samsung’s own radio units don’t cover. But there’s not a great deal of other information about the other O-RAN interfaces that are being developed. These other interfaces are the midhaul between CU and RU (F1), and to allow for Dual Connectivity of a device to equipment from different vendors (X2), for example.

In a way this mirrors a similar release from Nokia two weeks ago, where Nokia too announced a new vDU. At that point it said it supported O-RAN, but did not list out which interfaces. So we went to Nokia for the fuller answer and this is what it said.

“The X2 is specified in 3GPP and O-RAN defines specific interoperability profiles. Nokia X2 is already 3GPP compliant and Nokia plans the support of specific O-RAN profiles in the future based on customer requirements.”

That answer might have been a bit of a deliberate fudge so as not to prejudice an announcement it made this week in which it said it was “rapidly ramping up the adoption of Open RAN (O-RAN) interfaces in its AirScale portfolio”. In fact it said, “An initial set of O-RAN functionalities will become available this year, while the full suite of O-RAN-defined interfaces is expected to be available in 2021.”

That goes further than the reply to our enquiry two weeks ago, and seems to imply that Nokia is fully on board right across the range or O-RAN interfaces – although the vendor has recently queried whether tight coupling of certain functions – say the CU user plane and control plane – will make disaggregation achieveable and/or desireable. So that’s certainly one to look out for.

The wider question for strategists is, given that Nokia and Samsung are leading on O-RAN messaging amongst the larger vendors, are they heading to [some form of] Open RAN from positions of strength, and thereby ratcheting up pressure on Ericsson to do the same? Or is it in fact Samsung and Nokia who are under pressure and being forced to move to Open RAN in order to be part of current and future buying decisions? The two biggest selling 5G RAN vendors to date are, of course, Ericsson and Hauwei, neither of whom have conceded very far to Open RAN initiatives of various hues.

That’s not to say that they could not do it when they had to. Ericsson is involved in several O-RAN working groups and talks about vRAN as an ongoing development project with a clear, near-term commercial outcome.

This blog post from February points out that Ericsson sees integration of Open RAN as a key target. So it makes sense to see its position as one that it goes vRAN, with support for Open interfaces, when it judges the market is right to do so.

“Ericsson is actively working – independently and within the ecosystem – to explore COTS-based platforms that bring higher efficiency, and also with harmonization and standardization of vRAN components and architectures to minimize the cost of system integration. The target of this work is to make vRAN a commercially competitive solution that can complement existing investments in the near future.”

In the excitement about O-RAN and vendor diversity and politicians making statements about national security and all else, it’s tempting to think that those going open are necessarily those in the lead. But it’s not quite as simple as that.