The decision by five key vendors to forge common software interface standards for Open RAN begs more questions than it answers. The first question is why another Open RAN standards initiative is necessary at all. It prompts the suggestion that the standards bodies themselves are failing to do their job properly if these new interfaces are as essential as the O-RAN Alliance itself acknowledges.
Alternatively, it suggests that Open RAN is still very much work in progress and remains in a fluid state, where the areas for standardization are still being identified. In either case, it appears that Open RAN is not yet mature or stable enough for deployment by operators in public macro infrastructures, even if it can be in some closed private enterprise networks.
This certainly resonates with stated plans by those operators most positive about Open RAN, such as Vodafone, which has only earmarked 2030 as a target for converting 30% of its European sites to the approach. That would mean migrating about 30,000 4G and 5G cells to Open RAN by that year.
Vodafone was one of three tier 1 operators to welcome the new initiative called Multi-G from that quintet of vendors, Cohere Technologies, Intel, Juniper Networks, Mavenir and VMware, alongside Australia’s Telstra and Bell Canada. All three are customers of Cohere, which is spearheading this coalition, as a provider of spectrum multiplier software to all of them for trials.
Spectrum multiplication is increasingly seen as essential for optimizing use of what is a finite resource, exploiting the potential to increase capacity over a given frequency band with the help of massive MIMO creating simultaneous paths. To do this effectively, spatial multiplexing has to be employed to prevent interference between MIMO streams, and this can be done with an increasingly prevalent technique called Delay-Doppler.
By representing the channel geometrically across all user equipment (UE) attached to a base station, Delay-Doppler enables simultaneous orthogonal beams to be transmitted to multiple UE (User Equipment) devices over the same spectrum, effectively multiplying its capacity.
Cohere’s spectrum multiplier software can run on a near real-time RAN intelligent controller (RIC) platform as part of an Open RAN deployment, which is where this new standardization effort figures.
As its name suggests, the aim is to facilitate deployment of such Open RAN capabilities across multiple cellular generations, primarily 4G and 5G at present. This is what the five vendors claim is lacking at the moment, as reflected in their joint statement: “The Multi-G initiative would define frameworks, interfaces, interoperability testing, and evaluation criteria that would provide the interfaces to support full coexistence of 4G, 5G, and future waveforms. This effort would help drive higher performance and connectivity across satellite, private and ad-hoc networks, and autonomous vehicles, increasing new service and revenue opportunities for telecommunications and mobile operators.”
The standards bodies seem to agree with them, as Yago Tenorio, chairman of the Telecom Infra Project (TIP) and also director of network architecture at Vodafone, said in a statement: “We strongly endorse this initiative, and we look forward to seeing the critical interfaces published into the relevant O-RAN Alliance and TIP Working Groups.”
We put the obvious question to TIP, as well as to the O-RAN alliance, why if these interfaces are so critical it has been left to this new grouping to define only now. Both organizations declined to address the questions directly, with TIP merely supplying a rather elliptical quote from its head of comms Rudolph Moncrieff, who said: “Open RAN is ready to deploy today but work on standards affecting certain elements, like the RIC, are ongoing as they are still a work in progress despite the incredible promise shown to date.”
This suggests some unease over progress, and also perhaps a fear that the initiative might appear as a power grab by this group of vendors, or at least an attempt to wrest control over Open RAN development. Certainly, these vendors have articulated a fear that Open RAN might be trapped in a blind alley, and require rescuing to span multiple cellular generations within a more open multi-vendor framework.
There is another angle that is rather more positive, which is an attempt to unify the Open RAN movement behind all its constituent parts and end some confusion over terminology. There has certainly been confusion between Open RAN itself and virtualized RAN (vRAN) in the mistaken belief the two are synonymous, or at least sides of the same coin. While an open RAN requires virtualization, the reverse is not true in that a vRAN is not necessarily open.
It is true that the split RAN architecture originally defined by the 3GPP to enable URLLC (Ultra Reliable Low Latency Communication) and eMBB (Enhanced Mobile Broadband) set the stage for both Open RAN and vRAN. The O-RAN movement then set about defining interfaces between six critical components within this split RAN.
These are: Orchestrator and RAN or the A1 interface; RIC (RAN Intelligent Controller) and CU/DU (Centralized Unit/Distributed unit), the E2 Interface; CU-CP (control plane) and CU-UP (user plane), the E1 Interface; CU-DU, or F1 interface; DU-RU (radio unit), known as Open FrontHaul, and Orchestrator for Cloud Platform (O-Cloud), the O2 Interface.
For vRAN, a vendor would virtualize these RAN components such as BU, DU and CU, but this could be done with their own interfaces and does not have to comply with O-RAN specifications. It would still be virtual in that the component can scale independently by running on different hardware, but might not interoperate with equipment from other vendors.
However, the end game is to unite vRAN with O-RAN under common interfaces, such that operators can both combine components from different vendors and scale them independently, all within a common framework.
This explains the participation of Intel in the Multi-G initiative, given that its FlexRAN 4G and 5G baseband PHY Reference Design running on its Xeon series Processor is quite widely deployed in these RAN components. This is based on a native physical, or Layer 1, foundation but with proprietary controls on top. As part of the Multi-G initiative, Intel will rework FlexRAN around a common set of controls that will be defined by the body, as Cohere’s CEO Ray Dolan has confirmed will happen.
If this does lead to unification of O-RAN and vRAN, then the initiative will have achieved something. But there are still questions over fragmentation, given that Qualcomm’s position is unclear for example, again not responding to requests for clarification.
Qualcomm had announced a partnership with Dell to develop a virtualized distributed unit (vDU) system at MWC 2023. The two companies also pledged to combine Dell’s PowerEdge servers purpose-built for open telecom networks with Qualcomm X100 5G RAN Accelerator Card with commercial-grade Layer 1 software. This is clearly another land grab around the field, so there is still uncertainty over how the dust will settle around a common unified O-RAN and vRAN.