Perhaps new lockdowns and impending recession cast a shadow over optimism in every quarter at the start of 2021, but the discussions about open RAN certainly came with a hefty dose of weary scepticism compared to the enthusiasms of 2020. Would O-RAN be yet another example of a vision of resetting the rules in the wireless industry, that crashed on the rocks of entrenched interests and the practical challenges of deploying large-scale wireless networks?
There are plenty of past examples after all. To name just a few, we can remember the push for patent pools to standardize royalty payments in 4G handsets; the open equipment ecosystem that was supposed to accompany WiMAX; even the open source base station (which was conceptualized by executives from, of all companies, Huawei)?
The objectives behind these initiatives, and many others of the past two decades, had much in common with those of O-RAN (and more broadly, all open RAN activities). They aimed to lower barriers to entry for new vendors, make licensing lower cost and more transparent, provide a common platform for innovation, and give operators more choices and greater price competition. But they fell victim to lack of sufficient cross-industry support, especially from large stakeholders, and to in-fighting and fragmented agendas among operators and ecosystem players.
Just as discussion about O-RAN was starting to recall these sad episodes, a huge confidence boost was provided by the news of a memorandum of understanding (MoU), signed by four major Europe-headquartered operator groups (see below), in support of O-RAN. Deutsche Telekom, Orange, Telefónica and Vodafone have 1.4bn customers between them worldwide, so their scale is more than adequate to move the needle for a new architecture in terms of market confidence, revenue potential and deployment momentum. This makes up for the retreat of AT&T – one of the instigators of O-RAN – into a more cautious approach; the commercial challenges of open RAN pioneer Rakuten, which is rumored to be considering an exit from the mobile market; and the short list of tier one deployment plans that are not from greenfield operators.
This is a big step for O-RAN, and certainly cements its position as the current leader among various approaches to open RAN (in its broader sense), which also include initiatives by Small Cell Forum and Open Networking Foundation. But it does not guarantee commercial success for the architecture, and even less does it guarantee that the objectives of the movement will be achieved.
Support in MoU form from several large companies does not necessarily accelerate a market, though it certainly boosts confidence. Sometimes a big announcement by several lumbering telco giants can slow down a new platform or approach, because it introduces the politics and conflicting agendas, not to mention the committee approach to processes and procurement, that accompany any large organization.
This is not just true of notable failures like the operator-driven Wholesale Application Community (WAC). Arguably, lack of cohesive operator effort allowed large vendors to implement previous standard RAN interfaces, such as CPRI and OBSAI, in non-interoperable ways. While technologists from some operators bickered in committees, their deployment executives were looking for a solution that worked today, and the purity of open systems took a back seat.
The same could happen to O-RAN, if the operator MoUs, and the related moves of support by European and US governments, are confined to high level statements. Real, practical work and cooperation is needed between operators in a short timescale, and by people who understand the realities of at-scale deployment and optimization. That will translate into workable architectures – inevitably with some compromises on the full O-RAN dream – and will ensure that vendors provide support, however cautiously.
This does not guarantee a Nirvana with scores of vendor choices for operators. The large suppliers will retain genuine advantages, even if they support O-RAN enthusiastically, as Samsung is doing (for obvious reasons). Some innovators will be acquired by the big vendors, or by webscalers like Microsoft; others will not be able to take on the huge challenges of supporting a national macro RAN project and will specialize in the enterprise and private networks where new vendors and architectures will face far lower barriers.
In these latter environments, however, O-RAN itself may not be the best interface to form the basis of deployments. One of the warning signs over O-RAN is that, rather than providing operators with the flexible choice of different functional splits for different vRAN scenarios, it is forcing them into a single choice, Split 7.2. This entails significant challenges to deploy and afford the cloud infrastructure and fiber needed to achieve optimal performance in a high performance environment. It will be well-suited to some operator models, but in a more localized scenario, an alternative such as Split 6, favored by Small Cell Forum and its nFAPI architecture, may be more attractive.
Already, operators like BT – an early experimentalist with many open and virtualized platforms – are counselling caution. BT’s CTO, Howard Watson, recently said the idea of a macro RAN being procured from four or five different vendors was practically unrealistic. And senior BT executives are calling for a more flexible approach to open RAN, in which multiple splits, multihaul options and controllers can be accommodated in a software-defined manner.
These comments point to two risks for open RAN. Proponents are laying out a very ambitious vision, but operators need to start issuing RFPs and planning roll-outs very soon. So trade-offs must be accepted upfront, rather than risk disappointment and backlash. A multivendor ecosystem will provide a wider choice of suppliers, and easier swap-outs, but not a network where many vendors are integrated and re-integrated on a flexible basis. And the full vision of a software-defined, dynamic and open – let alone open source – RAN remains years, possibly decades, away, such are the changes it requires to procurement and deployment processes, pricing and licensing practices, testing and verification, ecosystem structures, and fundamental radio technology.
That full vision will not be delivered by O-RAN, but O-RAN can still have a significant impact, as long as large operators’ support is real and practical, rather than tick-boxing to pressurize existing suppliers and please governments. It can introduce significant new innovation and ideas. Arguably the most important contribution of the four European giants is not their MoU, or their promises to support O-RAN in a certain percentage of sites by 2025, but their incubation of many interesting RAN technology start-ups in recent years, sometimes under the auspices of the Telecom Infra Project (TIP).
We will see operators encourage their primary suppliers to include innovators in their ecosystems; together with increasingly strategic relationships with webscalers to enable vRAN, this will start to change the shape of the RAN market.
And we will see greenfield operators and enterprise or small cell deployers – which have far greater latitude to experiment with new architectures and vendors – blazing the trail and providing valuable case studies for the wider industry.
Samsung, which tried to ride some of those previous failed initiatives, such as WiMAX, to build a global RAN business, will succeed this time, if it can translate its US successes to other markets and address its lack of 2G/3G support (important in Europe, in particular). NEC has a strong opportunity to play a greater role in the radio/antenna space – and in the macro market, this will remain the segment with the highest walls for new challengers, as well as the primary obstacle to multivendor networks with RAN-neutral integration (an open RAN goal which seems unlikely to be achieved without accepting the radio vendors as the integrators).
The other group of suppliers to watch intently is the small band of semiconductor vendors. Qualcomm’s re-entry into the macro base station chip market last year was one of 2020’s most interesting developments, especially if the company can resurrect its aborted cloud processor efforts. It could join Intel, Marvell and potentially AMD (especially if it completes the acquisition of Xilinx) in building a complete platform for vRAN, from cloud processors to accelerators to radio chips. As equipment vendors are pushed away from ASICs in order to make the economics of cloud-based RAN work, the power will shift to the merchant chip providers.
Any of these chip giants – or a group of them if they choose to cooperate on certain specifications – could have a greater ability to define and drive common RAN platforms than higher layer efforts in which scores of operators and vendors try to form a consensus. The influence of large and small chip companies on making open platforms workable is already very clear in WiFi and small cells. That may extend to macro RANs in future too, though it threatens to create a new source of consolidated power and lock-in too, rather than the brave new world of a fully open playground where every kind of supplier can be on equal terms.
Recent O-RAN developments:
- VMware, a key partner in Dish Network’s planned open RAN deployment, is developing a lightweight version of its Telco Cloud software stack, that will be suited to the ‘far edge’ nodes on which highly distributed O-RAN architectures will run. This was initially designed for Dish but will now be productized for other customers.
Sachin Katti, VP of telco strategy at VMware, told TelecomTV: “You don’t need all of the features for RAN and far edge, so there is a version of the VMware telco cloud stack that will be specific to RAN deployments. That’s coming out in the next few months, but it’s already in preview with a few customers.”
He added: “So we are learning a lot about the automation and orchestrating requirements for the edge and far edge… you need to have a lot of zero touch capabilities. It’s kind of unique in the sense we are doing product development alongside the network deployment rather than saying, ‘Hey, here’s a product, go deploy’.”
- io is an example of a company that is riding the migration towards virtualized RAN to expand out of its data center automation market and into the telecoms world. It will provide application automation to support Rakuten’s move to fully cloud-native 5G networks and aims to productize the containerized orchestration layer to be easily deployable by other operators.
Fernando Cerioni, VP of product management at Robin.io, said in an interview: “Our solution goes beyond the virtualization layer to full automation. We’re running over 40 applications that are fully virtualized at Rakuten.”
Robin.io is targeting the private cellular market, where many open RAN deployments are expected to take off first. Here, it has announced a partnership with UK-based private networks and virtualized core provider, Quortus.
“The partnership between Quortus and Robin will enable the companies to offer a powerful cloud-native mobile core network that is fully automated, lightweight and ready to deploy in days, not weeks,” said Robin.io CEO Partha Seetala.
- The most obvious target market for open RAN in its early stages is for expansion into rural areas where low population density or ARPU makes the cost of a conventional build-out hard to justify. A recent example is Hotspot Network, a Nigerian service provider that is working with Parallel Wireless to install a 2G/3G/4G network to cover 2,000 villages.
The Hotspot project is working with Digital Farmers Club, leveraging Nigeria’s universal service provision fund to support smart agriculture. The Digital Farmers Club said it aims to create over 100,000 direct jobs and 1,500,000 indirect jobs across Nigeria.
Hotspot’s founder and CEO Morenikeji Aniye said: “Low ARPU in Africa makes the deployment of existing networks commercially challenging. Operators need to adopt a strategy that helps them focus on expanding the existing 2G or 3G networks and get ready to migrate to 4G and 5G at much lower TCO.”
Parallel is working with other African operators, including being part of MTN’s pledge to deploy 5,000 open RAN sites in South Africa and other markets.
- New US president Joe Biden has confirmed Jessica Rosenworcel as the new FCC chair, and assuming this is confirmed by the Senate, this will put a notably O-RAN-friendly person in charge of the telecoms regulator. Rosenworcel has been a strong advocate, in her time as an FCC commissioner, of net neutrality, expanded spectrum options and open RAN, including the potential for the USA to build a major wireless industry again.
In September she said in a speech: “Open RAN has extraordinary potential for our economy and national security. That combination is something to seize—especially right now in the early days of 5G wireless deployment. We must focus now on our competitiveness, on strengthening our alliances around the world, and on reasserting our values — by building a new market for 5G equipment. That is how we will restore American leadership and secure 5G.”
- Creating a wide ecosystem for open RAN will rely on software developers integrating their products tightly with multiple radio unit (RU) platforms. Mavenir has established a dedicated unit for this purpose, and is even planning to develop its own radios, as well as working with partners such as Taiwan’s MTI. Now Reliance Jio subsidiary Radisys has integrated its 5G NR Open RAN software with Benetel’s RU as well as signing a licensing deal for the Radisys 4G and 5G stacks.
The two vendors will offer a pre-integrated open RAN solution to help to accelerate deployments, especially for enterprise or private networks or smaller operators without large engineering teams.
Adrian O’Connor, CEO of Benetel, said: “Open RAN success will be linked to a close integration between a CU/DU solution and an O-RU provider. Radisys is the ideal partner for our solution given its leadership in the Open RAN ecosystem and its industry-leading LTE and 5G NR software stacks.”