There has been a lot of interest in cloud RAN amongst established MNOs in recent months with several of the biggest MNOs planning to launch fully virtualized commercial networks over the next two years and incumbent vendors such as Nokia, Samsung and even Ericsson planning to offer a suite of fully virtualized cloud RAN products.
Benefits of Cloud RAN
A virtualized and disaggregated cloud RAN network will enable MNOs to fully leverage the benefits of a 5G cloud-native architecture, which include scalability, network agility and automation, features essential to enable new services such as low-latency MEC services, massive IoT, etc. However, there are many challenges to overcome before MNOs start reaping these benefits. These include include a lack of fronthaul fiber capacity, performance issues, vendor lock-in concerns, network integration challenges plus an immature ecosystem.
An Emerging Ecosystem
The cloud RAN ecosystem is an ecosystem under development and it will take several years for it to mature and to offer a full range of products. Although a plethora of new telco-grade COTS servers have been announced by server vendors in recent months, a lot of the key technologies are still under development. For example, the market for hardware acceleration and network connectivity is still in its infancy. Counterpoint Research believes that a range of acceleration card options – both look-aside and in-line – and based on FPGAs, GPUs or ASICs/eASICS. will be required. In addition, network interface cards able to meet stringent timing and synchronization specifications are required for fronthaul connectivity. Ideally, these PCIe-based cards will need to be multi-function incorporating both network connectivity and hardware acceleration on a single card. No such cards exist today.
All cloud RAN deployments to date are based on COTS servers using Intel’s FlexRAN architecture with or without hardware acceleration. In fact, Intel’s FlexRAN is almost the only cloud RAN chip reference architecture available commercially today. However, there are numerous alternative platforms being developed by competitors. These include ARM-based alternatives from vendors such as Marvell and Broadcom, GPU-based designs from Nvidia as well as a few RISC-V-based possibilities from start-ups such as EdgeQ and Picocom. Counterpoint Research believes that the widespread commercial availability of these alternative merchant silicon designs will be critical to commercial development of the cloud RAN market.
Incumbent Versus Open Cloud RAN
Compared to their European brethren, major MNOs in the US seem less obsessed about swapping radios and basebands and are adopting a different strategy with respect to cloud RAN. While all major US MNOs agree that open RAN will be the future, in the short-term, they believe that the best way of benefiting from open RAN is to leverage its intelligent RAN capabilities, an inherent part of the O-RAN Alliance architecture, via the RAN Intelligent Controller (RIC) design. As a result, Counterpoint Research believes that most US MNOs will rely on cloud RAN solutions from their existing incumbent suppliers such as Ericsson, Nokia or Samsung, thereby avoiding the complexities of integrating open, multi-vendor networks.
Economics and Business Case
For established operators, the feasibility of deploying cloud RAN largely depends on three factors: the availability of cheap fiber, real estate sites (to accommodate data centers) plus a viable business case. Inevitably, this will vary from operator to operator and from country to country. Legacy telcos in countries with good fiber access such as Japan and South Korea who are able to adapt existing central office sites to accommodate cloud RAN servers are potentially at an advantage here. However, even for them the business case is not clear cut. At this time, the potential cost savings of centralizing and cloudifying the RAN remain largely unproven and hence Counterpoint Research doubts that the business case for deploying cloud RAN stacks up. Combining cloud RAN with MEC-type enterprise services may be a better bet although the MEC market itself is also a nascent and unproven market.
Cloud RAN Deployments
With China unlikely to go down the disaggregated route any time soon, Counterpoint Research believes that the options for large scale cloud RAN deployments seem limited in the short and medium term. And after spending billions of dollars, MNOs elsewhere are unlikely to rip out and replace existing 5G infrastructure. One opportunity though could be the US market with the deployment of new 5G networks in C-band starting in 2022. However, even here cloud RAN will probably be limited to a few small-scale commercial deployments (essentially trials) for the foreseeable future – at least until the ecosystem matures and the business case becomes clearer. In the meantime, big MNOs such as AT&T, NTT DoCoMo and Verizon (plus a few vendors) will continue striving to gain the upper hand in the all important PR game!