Open RAN products were present in force at last year’s MWC. Keith Dyer from The Mobile Network highlighted the appearance at last year’s MWC of a wide collection of emerging Open RAN Distributed Unit (DU) products [i].
These demonstrated progress amongst vendors in the softwarisation of many RAN functions, with the processing intensive Layer 1 functions present in the DU equipment element, as always, presenting a challenge and prompting much debate around the optimum choice of processing platform and companion hardware accelerators. Amongst the DU products shown last year it seemed to be nearly a given that softwarisation of the DU meant a transition to Intel x86 based platforms.
Fast forwarding to FYUZ 2023 at the end of last year a different debate around the implementation of DU products was gaining momentum. While the merits of alternate hardware acceleration approaches were still being argued, there was a notable presence from Arm and a promotion of the Arm ecosystem as a route to a diverse range of chipset options to fit the processing, form factor and power consumption requirements of DU product variants targeting different deployment settings.
At Real Wireless, we are looking forward to this year’s MWC and seeing how the next part of this debate on how to find the optimum choice of processing platform for Open RAN products unfolds. Inevitably, the processing intensive requirements of the DU and its interface to the RU will be the hardest and last remaining nut to crack to unblock the path to a fully abstracted cloud RAN with questions over how far we can get without companion hardware accelerators. Going a long way to answer many of these challenges, we are particularly pleased to see that the Parallel Wireless team will be at MWC exhibiting their hardware agnostic DU software solution which delivered an impressive world-first demonstration across an outdoor 5G private research network in Bristol at the end of last year.
Enabled by a grant from the Department of Science, Innovation and Technology (DSIT), the Bristol demonstration – enthusiastically reported by Iain Morris of Light Reading [ii] as a “cloud RAN breakthrough”- is an outcome of the Proteus project which productizes processor-agnostic RAN solutions, adaptable for Intel’s x86 as well as Arm-based technologies. This approach cleverly tackles the debate on optimum processing platform choice in a different way. It acknowledges that different chipsets have pros and cons in different deployment settings and that these pros and cons are constantly changing as silicon vendors continue to innovate and release new products. Therefore, enabling choice of processing platform, rather than picking the potentially short-term “optimum” chipset, is at the heart of the Proteus project.
Notably, the Proteus solution makes no use of hardware accelerators taking the industry a step closer to fully cloudified RAN solutions for the deployments that merit it, such as the cloud-based private 5G network as a service solutions being considered by some for delivery from hyperscaler platforms.
Real Wireless is proud to have been part of the Proteus project, alongside Parallel Wireless, University of Bristol, BT Group, Arm®, NXP® Semiconductors, AttoCore, Benetel and CableFree. Our role has been to quantify the benefits that having more choice in processing platforms for Open RAN products might bring to vendors, MNOs and the UK’s ambition of accelerating the market adoption of Open RAN solutions.
Our Proteus contributions have included the development of a model to dimension the processing requirements of DU product variants targeting different cell site types. We then map these processing requirements to suitable processor and server products available for the Intel x86 and Arm-based processing architectures the project targets. Model outputs include a bill of materials (BOM) for potential Open RAN product variants which in turn can be translated into CAPEX, power consumption and OPEX comparisons to analyse the potential for power consumption and TCO savings enabled by choice of processing platform.
Our Proteus benefits analysis was performed for different deployment types such as coverage focused rural cell sites and capacity focused urban cell sites for both cloud native and bare metal solution to understand the merits of different processing platforms in different settings or product variants. Our results have highlighted the value of the choice of processing platform and helped unpick some of the details around the merits of one approach over another in the ongoing debate surrounding the implementation of Open RAN products and the transition to fully abstracted cloud RAN. They also highlight how a more diverse Open RAN value chain has the potential to contribute to faster Open RAN adoption.
This type of analysis is clearly helpful for chipset vendors and mobile equipment vendors deciding on Open RAN strategies and product roadmaps. However, it also has implications for wider infrastructure deployment decisions for MNOs, neutral hosts, and other infrastructure providers (including hyperscalers and data centre providers). It is intrinsically linked with the debate on the pace of transition to virtualisation in mobile networks and even network slicing – all areas that we have been working with tier 1 vendors and MNOs on for nearly 10 years in our participation in European Commission sponsored projects.
At MWC 2024, it is clear that the debate on “what good looks like” for Open RAN products and Open RAN deployment strategies will continue. We look forward to continuing to play a key role in this debate by helping our clients quantify, evidence and understand the trade-offs to navigate towards their optimum solution.
