Part I: O-RAN Alliance
Are operators really serious about putting large chunks of their RAN business the way of non traditional vendors that build to open interfaces within decomposed architectures, with virtualised radio software deployed on commercial grade hardware?
Although TIP’s Open RAN programme, the O-RAN Alliance and other efforts such as Cisco’s Open vRAN ecosystem to build open RAN ecosystems have been well publicised and done much work, there has been scepticism that operators are really going to take the jump and trust their radio network to run as virtualised RAN processing software instances on white box hardware.
Andre Feutsch, Chair of the O-RAN Alliance, President of AT&T Labs and AT&T’s CTO, said simply that they will. He told a meeting at MWC that the open RAN – with its disaggregation of RAN functions – is not “rocket science”. The process mirrors what the industry has achieved in routing and switching and L4 type services.
“Follow the money,” he said – pointing to the 20 operator members of O-RAN and to what he said were their considerable investments to date. Alex Choi, SVP Innovation Deutsche Telekom, said the group had 160+ people at its meeting in Madrid 2 days before the start of MWC, indicating “how relevant this is to whole ecosystem”. The Alliance had six O-RAN based showcases spread out over a dozen booths.
These included demos by NTT DoCoMo, Keysight, Mavenir and NEC of Baseband and Radio Units from different vendors interoperating over the Open Fronthaul interface, to an implementation by China Mobile and ZTE of the Radio Intelligent Controller load balancing UE attachment, to a virtualised Central Unit on an Akraino Edge Cloud on the Nokia stand. Xilinx was showing a programmable mmWave whitebox Radio Unit, while China Mobile, Intel and Lenovo had a live demo of whitebox hardware supporting sub 6GHz indoor coverage.
From showcases to deployment
So how does Feutsch see operators introducing open RAN networks? Where will we see them deploy the first instances and how will they deploy? He said that one area that might lead O-RAN investment is indoor and enterprise use cases, with the enterprises themselves acting as a customer base in some cases. “We might see certain use cases where not only operators want to use it but enterprises get involved – and that could create volumes and scales for these in-building solutions.”
Feutsch said O-RAN Alliance members want to achieve a genuine disruption to the market. They’re not playing games. “The traditional ecosystem is ripe for disruption,” he continued. “Take a look at the spectrum coming on line that can be taken advantage of. This is about bringing more innovation into the ecosystem. It’s the power of the many versus the control of the few. I see a lot optimism that this will work.”
Feutsch didn’t say it but you could also follow the direction of travel of the traditional vendors most threatened by a disrupted RAN. Nokia has been a contributor and member to O-RAN all along and sees the benefit of working in an open environment. This appears to be a contradiction to its oft stated advice that using Nokia “end to end” for 5G, including its dedicated Reef Shark chips, delivers the greatest operational advantages for operators. But the vendor is also canny enough to know that there are few end to end contracts up for grabs. So being open means that it can play both ways. It can plug in as a vendor where an operator is building a multi-vendor strategy, and also be a trusted integrator and provider of other services (automation etc) within that wider network. For example, it is unlikely that Nokia would have the role it has within Rakuten’s “all cloud, all virtual” new network build, where it is not the vRAN supplier, if it had not adopted this strategy.
Ericsson has just joined O-RAN, albeit slightly hesitatingly in terms of endorsing the whole scope of the project. But it is keen to become more involved in upper layer automation, and its public adoption of Intel’s new 10nm Snow Ridge platform to act as a boost to RAN compute should not go un-noticed here either.
Huawei has been much more publicly sceptical about the open RAN model – stating that Intel CPUs are not up to the job of running a RAN, that equipment requires 10x more power and performs worse than its own, hardware-optimised platforms. It has not joined O-RAN.
Chih-Lin I, Chief Scientist, China Mobile Research Institute, said there has been some misunderstanding on that front. Not everything will run solely on x86 processors, even in an open RAN. There will be specific accelerators where required. Feutsch added that there will “of course” be DSP requirements; open RAN doesn’t necessarily mean you can put everything on a simple generic processor and it will all work.
Chih-Lin said we could see many diff versions of a white box, some using ARM or X86 but some also being highly hardware-centric. A simple reference design could provide a starting point for maybe Tier two/three solution provider, with everyone else adding their “secret sauce” to differentiate. That helps the ecosystem have a lower entry barrier.
Caroline Chan, VP and GM, 5G Infrastructure Division, Network Platform Group, Intel, said that there would be different models coming to market, including indoor small cell versions. There would also be new entrants that are not currently telecoms suppliers. “There are non believers, but the action the group has taken, the proof points we have produced, we have started showing why open interfaces and disaggregation do drive the pace of innovation. The more open the system, you attract a different type of ecosystem – those that don’t participate in telco today.”