Open Conflict Over Open RAN
Strange as it might seem, Huawei's international tribulations may bear some responsibility for the growing interest in a more open radio access network (RAN).
As security hysteria drives countries to excommunicate the once-indomitable Chinese company, telecom operators are left with fewer alternatives than ever before. Their dependence on a shrinking number of industry titans has made them push even harder for more "open" technologies -- and especially for a radio access network (RAN) without a dominant supplier. (See DT's Dinner With Huawei Could Become Dog's Breakfast and Major Telcos Pool Efforts to Slash 5G RAN Costs.)
Some form of open RAN could soon become a reality. Yet its genesis has been turbulent, and its outlook is fraught with uncertainty. Critics with an open RAN agenda say Huawei Technologies Co. Ltd. and Ericsson AB (Nasdaq: ERIC), the world's biggest mobile kit vendors, tried to smother it at birth, nursing an alternative that promised far less freedom.
Ericsson, for its part, says focusing on other 5G technologies was deemed more important at the time, and insists it is no enemy of the open. But while the Swedish vendor has now enrolled in the main club backing open RAN developments, Huawei remains conspicuous by its absence. Even with the support of the world's largest operators, can an open RAN thrive amid such conflict and competing interests?
A brief history of open RAN
Today, the RAN is not only one of the costliest parts of a telco network but also one of its most restrictive. Although any RAN should talk to any smartphone, or to any transport network, the major components of a RAN must come from the same vendor's system. Buy signal-processing equipment from Ericsson and you also need its radios, or those of an Ericsson partner. Trying to build a RAN from different vendor systems is like piecing together a jigsaw from incompatible sets. It simply won't work.
The blame for this situation falls heavily on a maligned technology called CPRI (which stands for common public radio interface). It is CPRI that handles the connections between the signal-processing equipment (called baseband, in the industry) and the radios in what operators call the "fronthaul" part of the network. But there are so many variables with any CPRI implementation that operators are forced to work in one vendor's system. It is effectively a proprietary tool.
Ending this "lock-in" meant coming up with an open alternative to CPRI that would accommodate new demands for network design. In a typical 4G network, the baseband and the radios both live at the towers, with CPRI connecting them over a short distance. Rehousing the baseband in centralized facilities could bring benefits in future. But moving network functions into those centralized baseband units is far from straightforward, largely because of the increased fronthaul distance.
This "functional split," determining which functions go into baseband and which should be done at the radio site, has been at the crux of the open RAN debate. After considering a menu of splits, standards bodies like the European Telecommunications Standards Institute (ETSI) and the 3rd Generation Partnership Project (3GPP) had settled on two, according to John Baker, senior vice president of business development for Mavenir, a software company hoping to profit from a more open RAN. At opposite ends of the split spectrum, each had its pros and cons. (See Is vRAN Still Too Hot to Handle?)
The first, Option 2, would push many of the network functions into the radio unit. While this higher-layer split might help to reduce transport costs, it could also increase the signaling delay, making it unsuitable for latency-sensitive applications such as gaming and virtual reality. With Option 7, the lower-layer split under serious consideration, nearly all functions would shift into the baseband. That should assist with resource allocation to users, and boost performance, but it would also put more pressure on the fronthaul.
The most ardent supporters of open RAN saw reasons to favor Option 7, according to Baker. By leaving hardly any functionality in the radios, this would simplify the hardware requirements and help cultivate a market for "white box" radios -- standardized kit that operators could buy "off the shelf," like tools at a DIY store. Option 2, by contrast, seemed to reinforce the vendor lock-in. "It puts a lot of functionality into the radios and so they are still proprietary," says Baker.
Next page: It's just not cricket
It's just not cricket
But Ericsson and Huawei refused to play ball, he says. They naturally viewed Option 7 as a threat to their market dominance and did not want to embrace it. "Ericsson stopped the whole discussion about Option 7 in the 3GPP and totally blocked it," says Baker. "Ericsson and Huawei controlled what went into the 3GPP."
Alain Mourad, a director of engineering at R&D specialist InterDigital Inc. (Nasdaq: IDCC), is equally scathing. "This is what the vendors did not want to happen because they would be losing in business," he told Light Reading during a recent company event attended by analysts and reporters. "So what they have done is say we'll move a bit from CPRI but emphasize those [split] options that would require an additional box."
Moving a bit from CPRI, as Mourad puts it, meant designing a successor called E-CPRI. But that is not radically different from CPRI, says the InterDigital man, and certainly not about to usher in new RAN competition.
Of course, neither Mavenir Systems Inc. nor InterDigital is an impartial observer. Mavenir is trying to establish itself in the RAN as an open alternative to Ericsson and Huawei. Long accused of being a patent troll, InterDigital has previously fought Ericsson over intellectual property. Separately, it was this week reported by the Economist to have been "mauled" in royalty-related antitrust disputes in China.
Ericsson does not deny that standards bodies decided not to proceed with the option Mavenir favored. But it says there were good reasons behind that decision. "It was the conclusion of the 3GPP working group at the time that they wanted 5G to be open to different identified lower-layer split options, and that the detailed standardization of this interface targeting one split is limiting, since it will lock down the functionality for an area (active antenna systems) that is currently undergoing large innovation and development," it said in a statement emailed to Light Reading. "Hence, there is a risk that this leads to a sub-optimal and costlier 5G system."
As a result, the 3GPP agreed to park this work item, Ericsson explains, and focus on the 5G new radio (NR) interface and core network options. Specifications for NR technology were frozen at the end of 2017 and commercial 5G services are now appearing -- much sooner than anyone had originally expected.
Huawei was approached by Light Reading but at the time of publication had not responded to the criticisms leveled by Mavenir and InterDigital or answered questions about its interest in open RAN developments.
Nevertheless, several companies, including some of the world's biggest telcos, did go outside the 3GPP to work on a CPRI alternative. Formed in late 2016, the xRAN Forum published its first specification in early 2018, shortly before it merged with China's C-RAN Alliance to become today's O-RAN Alliance. It now includes at least one Tier 1 operator from just about every part of the world.
Baker, who has been heavily involved with the development and promotion of the xRAN specification, says it tackles the ambiguities found in CPRI and should allow one vendor's baseband gear to function with another's radios. "Mavenir has its own BBU [baseband unit] and more than ten partners working on white box radios," he says. His prediction is that white boxes could eventually account for between 20% and 30% of the entire RAN market.
Next page: Resistance may be futile
Resistance may be futile
Yet not all the industry giants have been resistant, says Baker. A member of the xRAN Forum before it morphed into the O-RAN Alliance, Nokia Corp. (NYSE: NOK) has been a "good contributor" to specifications, he says. In his view, the Finnish equipment vendor has simply been "more intelligent" than Ericsson or Huawei in responding to new customer demands. Even if the open RAN does lead to a fall in sales, Nokia may have reasoned that a stubborn refusal to move with the times would be far worse. (See Why Resistance to the Open RAN May Crumble.)
Now the xRAN specifications are a fait accompli, others are starting to budge. Ericsson joined the O-RAN Alliance earlier this month in a move that may have seemed counterintuitive to critics. But Erik Ekudden, its chief technology officer, denies Ericsson's strategy is not aligned with the goals of the O-RAN Alliance. Inside the group, a focus area for the Swedish vendor will be getting the RAN to communicate automatically with different network management systems. Providing this support could be a future business opportunity for Ericsson, he insists. (See Why Ericsson Is Joining the ORAN Alliance and Ericsson Weighs ORAN Alliance Membership.)
Its enthusiasm for the xRAN specifications is less obvious. Ekudden dismisses any suggestion these will eventually replace CPRI and E-CPRI. "There is no fast change in that respect. E-CPRI is already maturing," he tells Light Reading. "There is discussion in O-RAN about novel deployment options that would take part of that space because the existing [CPRI] specs are performing very well for tasks already today."
Gabriel Brown, a principal analyst with the Heavy Reading market research business, dismisses the charge Ericsson is trying to crush the open RAN as nonsensical. "In the O-RAN Alliance, it does appear that Ericsson is focused on the higher-layer split and RAN automation, but this doesn't preclude other parties from following another path -- that's the benefit of an ecosystem," he says. A new entrant, he points out, will naturally have a different perspective from the "larger, systemically important vendors." Moreover, an open RAN is about more than just a fronthaul interface. "A closed baseband-radio interface is not why the RAN equipment market has consolidated," says Brown. "There are many factors behind this."
Brown has also previously argued that partnerships between larger vendors and specialist radio developers could be mutually beneficial. "There is a lot of innovation in radio products, but the process of OEM [original equipment maker] integration slows down the rate at which this innovation can be deployed," he told Light Reading during an interview last year.
Even with the midwifery of Ericsson, Huawei and Nokia, a mass market for white box radios could have a difficult birth. It is not just the dug-in position of CPRI that matters. As operators have realized with the transition to software-based networks, working in a truly multivendor environment can be an operational headache. In a Heavy Reading survey of telcos carried out in December, expertise in systems integration ranked as the number-one requirement for an open RAN. "You need to match openness with stability," says Brown. "Realistically, which operator is going to deploy in a mix-and-match way? Probably very few."
Fronthaul restrictions might also continue to rule out Option 7. Baker champions the xRAN specifications based on this split as a "macro cell" technology, but he says the O-RAN Alliance has been careful to accommodate Option 2 in its work. "Option 7 is if you know the latency of the network you are designing, whereas if you have unknown latencies then Option 2 is more tolerant," he explains. "You need both in the toolkit."
The open RAN might seem to be the last of Huawei's concerns right now. Meng Wanzhou, its chief financial officer, is being held in Canada on charges of fraud. A ban on the sale of US components to Huawei is now a possibility. And several countries have banished it from equipment markets. But as conditions worsen, the calls for a more open RAN are likely to grow louder, piling further pressure onto the beleaguered Chinese vendor. (See Huawei Hits Out at DoJ Amid Global Backlash.)
"It [Huawei] must be interested in what's going on, but so far its position is that the market for open RAN is small and that the performance of these systems will be inferior to its own integrated product," says Brown.
As for Ericsson, its commitment to openness will be under renewed scrutiny now it has finally joined the O-RAN Alliance. One area where it says it plans to contribute is in developing interfaces between the RAN controller and the core network. "That is the other sticky area in terms of multivendor networks and how they interoperate," says Baker. "It has as much to lose there as on the radio interface." But as telcos fight back against years of vendor consolidation, the consequences of resistance could ultimately be far worse.
— Iain Morris, International Editor, Light Reading