One of the hottest topics in mobile networks is currently Open RAN. An access network in which elements from different vendors can be easily mixed and matched, and swapped in and out, could transform the economics of deployments. No more vendor lock-ins, and the flexibility to select the most advanced, or cheapest, component at any one time.
But ‘open’ can very easily become ‘fragmented’ and even ‘chaotic’. Combining components from different suppliers relies on standard interfaces to link them together in a common way. If there are too many of these interfaces, the industry will be split, and will lose the global scalability that uniform standards brings. But if a single interface is too rigid, it will not support the many different architectural choices that operators may make.
Common interfaces have been at the heart of Small Cell Forum’s work since, a decade ago, it defined Iuh, a femtocell interface that went on to be standardized by 3GPP. It is logical, then, that SCF should be in the forefront of developing specifications for a new wave of deployments, since many of the early activities in opening up the RAN are happening first in the small cell layer.
The key to the dream of the open network lies in disaggregation – separating out the key elements such as centralized units (CUs) and distributed units (DUs); and then open reconfiguration – combining components from any suppliers because they are all interconnected in the same way. For 5G, those central processes will usually be virtualized (run as software on off-the-shelf servers).
There are two potential barriers, however. There needs to be agreement on where the network should be split between the central and the local elements. There are 10 different options outlined by 3GPP, for instance. And there needs to be a single common interface between the elements in each split, or preferably one that can adapt to support any split.
There are many industry initiatives around the Open RAN issue. Each is valuable for engaging different stakeholders and adding to the wide range of ideas and innovations under discussion. The Linux Foundation’s ORAN Alliance, the Telecom Infra Project and the Open Networking Foundation’s CORD project are just a few examples. But there will need to be cooperation and consensus between different groups in future, to avoid fragmentation around many solutions – a situation which would see many MNOs losing faith in the open network and reverting to single-vendor solutions.
The impact on the economics of densification and 5G would be very high if that happened. The issue has been highlighted by the difficulty that operators find when they introduce 5G New Radio Non-Standalone (NR NSA) to their networks. Since NSA still uses the existing 4G core, it is very difficult to add 5G equipment from a different supplier from the 4G system, because crucial interfaces like X2 are implemented differently by each vendor.
These migration challenges are redoubling operators’ determination to support fully unified interfaces for their next deployments, and the best place to start is often in the relatively greenfield roll-outs of small cells.
This is why the Forum’s work on its 5G FAPI family of interfaces is so important. It will support multivendor interoperability between different components at different layers of the network, reflecting different 3GPP functional splits. As it evolves, it will have the flexibility to support different splits, and so improve the future-proofing of small cell deployments. In turn, that will bolster operators’ confidence to densify at scale and to involve a wider range of suppliers.
The first 5G FAPI (functional applications programming interface) extends the original 3G and LTE specifications, which are incorporated in most small cell chips today. It defines internal interfaces between the two chip-level network layers, the PHY and the MAC (medium access control), also known as Layers 1 and 2.
This is a truly universal spec. The L1/L2 interface is a pragmatic one, reflecting how subsystems are designed separately by different specialists, and then combined by a system vendor. The PHY API applies to all functional split options, and may be contained within the CU or DU components.
Future members of the 5G FAPI family will support more disaggregated architectures in which operators will need to make decisions about how they split their centralized and distributed resources. For instance, SCF will develop a 5G version of its Networked FAPI spec, which will enable a split MAC and PHY in a disaggregated small cell network, supporting the 3GPP Option 6 split over Ethernet fronthaul and targeting, in particular, cost-effective indoor scenarios.
These disaggregated networks will be harder to engineer than architectures where the CU and DUs are collocated but they will become increasingly important, and in future there will be dynamic splits, in which the balance between centralized and distributed network resources is altered according to the needs of a particular service.
There is still uncertainty among operators about how disaggregated the networks will be in future, and the best splits that will balance the low latency of a very distributed architecture with the efficiency and scale of a centralized one. With so much undecided, flexibility and openness are critical to avoid technology dead ends.
At the Small Cell World Summit in London in May, speakers expressed concern that there are still too many options and that MNOs either do not know which one they want, or each wants a different one, which limits scalability.
BT spoke for many operators when it cautioned that we do not yet know the extent of the benefits of disaggregation in real deployments and these must be traded against the additional costs of meeting tougher transport requirements. Its ideal scenario is the development of a platform which would support all the split options in software, to avoid the need to choose just one.
Such debates are important to crystallize operators’ thinking about the benefits and trade-offs of full disaggregation, and many will not finalize their 5G architectural strategies for some years.
However, what is very clear is that their decision processes will be made far easier if they have flexible solutions that do not tie them to a single option, and if the key interfaces are fully open and common. The Forum’s FAPI work, and the fact that many at-scale small cell deployments are greenfield and so open to new approaches, are two important factors in the push for the Open RAN. Together, they mean that the fastest progress towards a fully open network is likely to come in small cells, and that will help to accelerate densification in the early years of 5G.
To find out more about 5G FAPI or download the specification view the FAPI landing page here.