Your browser is not supported. Please update it.

5 July 2018

Open RAN platforms move closer to reality, but no common network in sight

Mobile operators are becoming more collaborative as they seek to define their own open platforms, rather than have architectures imposed on them by vendors. But this does not mean that the process of creating common frameworks is a harmonious one. Large operators have all kinds of conflicts of interest with one another, as well as different viewpoints on the best approach to a virtualized, software-defined 5G network.

So, rival operator-driven alliances are emerging in several key areas of the new network, including the MANO (management and orchestration) layer – where the options are led by ETSI Open Source MANO and the AT&T-driven ONAP – and the RAN itself. In the latter, another AT&T-originated project, now called Open RAN (ORAN) Alliance, will duke it out with the OpenRAN Project Group within Facebook’s Telecom Infra Project (TIP), Cisco’s Open vRAN, and elements of other programs such as CORD (see previous item).

That could create fragmentation, if no single initiative emerges with the highest level of support. If several prevail – which may become further split, by the nature of open source innovation – vendors will have to make difficult choices of which, and how many, to support. Ironically, that could play against the operator-supported start-ups like Orange-backed Athonet, since large vendors will be better placed, despite their distressed economics, to support several architectures and please as many MNOs as possible – and to reassert their power by working on interfaces and super-orchestrators to allow all the platforms to interwork.

Last week saw two of the open RAN projects making significant steps towards adoption by major MNOs. Virtualized RAN architectures have been developing for some years, as inhouse operator projects at China Mobile, KT and others, or within conventional standards bodies like 3GPP. However, over the past year, the open initiatives have turned their attention to the network element which will be most challenging to virtualize (partly because part of it, at the cell site, must always remain physical), but which could have the biggest impact on capex and opex reduction.

Within TIP, the OpenRAN Project Group has marked a significant step towards commercial reality, with Vodafone and Telefónica  both issuing DFIs (requests for information) from vendors to evaluate how they would support the OpenRAN requirements. The results of the RFIs, and the vendors which meet the requirements, will be announced at the TIP Summit in London in October.

“The vendor technologies presented in response to these RFIs will be evaluated using clear benchmarks that will help guide operators towards the selection of a shortlist of vendors,” according to a TIP blog. “Respondents are invited to submit comprehensive information describing the current state of their products and development plans for the next year. Only RAN nodes which demonstrate open interoperability from the perspective of the baseband processing platform, radio hardware, software and business model will be evaluated within this process.”

In one way, this looks like a radical open source approach being pulled back towards telco norms. Already, TIP has had to introduce a paid licensing process, in addition to its open source activity, to enable manufacturers of new-style equipment to participate. It’s one thing to offer software freely and then build a services model around that in order to monetize; that becomes a lot harder when base stations are involved and some TIP start-ups had said they were unwilling to contribute the inventions which were their key differentiator.

Now we see big operators issuing RFIs in the traditional way, presumably to big OEMs (among others). An RFI is often a precursor to an RFP (request for proposal), and the telcos’ RFP processes are notorious for requiring significant resources just to respond.

So it remains to be seen whether initiatives like TIP OpenRAN will end up with limited impact – an interesting way for operators to share ideas and tap into innovation in the broader ecosystem, but in the end, still relying on the usual vendors to turn MNO requirements into commercial systems.

Or this could be the first step in something more disruptive. For instance, new vendors may be announced as meeting the requirements, which would be a step to fulfilling the goal of many of TIP’s operator supporters, of shaking up the supply chain and introducing new providers alongside, or even instead of, the giants. And as the blog states, no player will be considered if it does not demonstrate interoperability.

The move to an RFI is positive because it shows that, after about a year, the OpenRAN group has achieved concrete requirements which are sufficiently unified and specific to be put out to the market. That will counter some critics who say open initiatives are little more than talking shops, which do not fit into the telcos’ established procurement processes.

And we have to remember that, before OpenRAN was set up, TIP had already helped to set new rules and expectations for the future mobile network. Facebook, for instance, had already contributed its OpenCellular small cell and Terragraph 60 GHz backhaul designs to TIP, providing specifications for vendors to build low cost, open access points, MIMO antennas and millimeter wave radios. This was all about commoditizing hardware elements which have traditionally be expensive and proprietary, and OpenRAN will complement that by working on the virtualized, software aspects.

The RFIs will also put pressure on the large vendors to make some difficult decisions. Will they accept these operator specifications, even though they are likely to undermine their traditional economics and architectures? Or will they try to spread FUD (fear uncertainty and doubt) about unproven architectures and the perils of open source fragmentation and security, and hope the fad goes away?

It will be a gamble either way, but eventually, it will be in the MNOs’ interest to get the big data center and network equipment vendors on board to bring scale to the market, and surround the new platform with deep knowledge of the telco business and legacy systems.

The relatively quick move to the RFI stage, while probably designed to put pressure on those large vendors and flush them out, also reflects the very pragmatic approach of TIP OpenRAN. Its focus is not on defining grand new architectures, but on getting hardware and software into the commercial space quickly, to start broadening the supply chain and allowing operators to harness commoditized compute hardware. This is in contrast to the ORAN Forum and Open vRAN, which are also developing specifications and network architecture.

Santiago Tenorio, Vodafone’s group head of networks strategy and architecture, said the RFIs reflect “strong industry collaboration via the TIP community,” and that it should serve to “help align the different open initiatives we have seen flourishing recently” by making it very clear to vendors what MNOs require.

What they require centers on decoupling the programmable RAN software from the hardware, and then splitting the control plane and user plane, adding flexibility and resource efficiency to the network. But instead of designing that virtualized, decomposed architecture from scratch, OpenRAN is drawing up agreed requirements, which could then be met, in theory, by a range of approaches. Nor, unlike ORAN Alliance, is it only focused on 5G – 2G, 3G and 4G integration or expansion are also covered.

Deutsche Telekom has said an open RAN architecture like this should reduce its 5G capex bill by at least 50% compared to 4G and has put its money where its mouth is, by being a founder member of the xRAN Forum, and by incubating several disruptive network equipment start-ups under the auspices of TIP.

“We are committed to the evaluation, development, and deployment of OpenRAN,” David Del Val, Telefónica’s I+D CEO, said in the TIP press release. “This RFI is a clear step towards commercialization of these new solutions that will simplify cellular networks deployment and operations, and unleash innovation for a better service to our customers.”

Meanwhile, the ORAN Alliance – a merger of the AT&T-founded xRAN Forum and the older Cloud-RAN Alliance – has added seven major operators to its list of supporters. When the merger was announced in February, the founders of both groups formed the core body of operators, consisting of AT&T, China Mobile, Deutsche Telekom, NTT Docomo and Orange. Now they are joined by Bharti Airtel of India, China Telecom, KT and SK Telecom of Korea, SingTel, Telefónica and Australia’s Telstra.

It also took contributions from Nokia, NEC and Samsung for the first set of xRAN specs. Other participants include C-RAN specialists AltioStar and ASOCs, plus Amdocs, Aricent, Blue Danube, Ciena, Cisco, CommScope, Fujitsu, Intel, Mavenir, Netsia, Radisys, Stanford University, Texas Instruments and University of Sydney.

With many of the world’s largest operators signed up, it will be increasingly hard for large vendors to ignore – and indeed Nokia, the most progressive of the big three in terms of the software-driven network, is already an xRAN Forum member.

The Alliance will examine many aspects of a software-defined, interoperable architecture, with fronthaul – the link between virtualized baseband units and the remote radio heads – being an early focus. The group has inherited work already started by the xRAN Forum on fronthaul, defining an interface which could be an alternative to the vendor-controlled CPRI.

“The specification exists and is fully defined in terms of protocols, and the O&M [operations and maintenance] piece is being added,” said John Baker, an SVP at xRAN Forum member Mavenir. “It is based on IP transport instead of expensive proprietary hardware.”

Another interesting question is how far these open initiatives will end up within the traditional standards process. Pure open source projects find it hard to marry their processes with those of bodies like 3GPP, though they often seek to interwork – and ETSI has been steering a delicate path between the two approaches. But many operators will not, in the end, feel comfortable adopting standards that are sanctioned only by an open initiative, and will want the certainty, and the certification processes, of 3GPP. So ORAN Alliance, though hosted by the Linux Foundation, has already spoken of trying to get its fronthaul interface specification adopted by 3GPP – which would also give it a significant edge over the several other fronthaul initiatives out there.

As well as the new members, the ORAN Alliance named Andre Fuetsch, the president of AT&T Labs, as its chair and Alex Choi, SVP of strategy and technology innovation at Deutsche Telekom, as operations officer.

It has set up seven working groups, which will focus on:
• Further development of the open fronthaul interface.
• Use cases and architecture.
• The A1 interface.
• The E2 interface.
• A stack reference design along with the F1/V1/E1/X2 interface.
• Cloudification and orchestration.
• White box radios, which could be purchased off the shelf and used with any vendor’s RAN software.

This represents rapid progress since the xRAN Forum provided its last major update in January, when it released its first open source specs, before announcing the planned merger with the Cloud-RAN Alliance. Then, it said that, within its remit to decouple the RAN control plane from the user plane, it had identified three key areas of effort to help to transform the RAN:

• standardized northbound and southbound interfaces to make it easier to deploy services and achieve multivendor networks, in a decoupled, software-defined environment.
• implementation of RAN software and network functions on off-the-shelf hardware,including the modular xRAN LTE eNode B base station software.
• and the flexibility for operators to run software functions at the macro cell site or at the edge of the network, depending on availability of fronthaul links.

Antenna vendor CommScope has already announced a 5G radio/antenna solution based on the xRAN interface specifications, which takes instructions from a virtualized baseband and integrates a beamforming active antenna array operating in the 28 GHz millimeter wave band. Initially, it is targeted at fixed wireless access (FWA).

“The release of the xRAN Fronthaul Specification is a groundbreaking advancement toward enabling an open RAN architecture to support next generation products and services,” said Bill Stone, VP of network technology development and planning at Verizon. “xRAN compliant radios coupled with virtualized basebands provide much needed flexibility to support rapid development and deployment of RAN products. By adopting xRAN specifications, we will be able to speed innovation, increase collaboration and be more agile to a quickly evolving market.”

Some large vendors are biting the bullet, perhaps to wrongfoot even larger rivals. Samsung said it played an “instrumental role” in defining the user plane, control plane and synchronization specs for the xRAN Fronthaul Specification Version 1.0. Alok Shah, VP of networks strategy, business development and marketing at Samsung Electronics America, told FierceWirelessTech: “Operators need new tools in order to manage continued growth in mobile data usage, and an Open RAN architecture can bring accelerated introduction of new features and technologies, greater flexibility in network architecture and operation and lower operating costs. Samsung views carrier grade open source initiatives as an important part of the move to network virtualization and automation.”

And Nokia also contributed to the specs, while earlier this year, it announced a deal with NTT Docomo to provide virtualized baseband units that could work with non-Nokia radios. This may be a way for the Japanese operator to introduce a new baseband while retaining the equipment it secures from local suppliers NEC and Fujitsu (which work heavily to Docomo’s specs as they have little position in the RAN market outside Japan). But it shows Nokia’s new willingness to be more flexible in the systems it sells to major clients.