It’s a race to the cloud as operators embrace network function virtualization (NFV). Previously, NFV had been largely centered on the core rather than the access network, but things are now about to change.
The focus of virtualization is starting to shift to the Radio Access Networks (RAN). This means operators are preparing and planning for the emergence of virtual RAN (vRAN) and its evolution to Cloud RAN with technologies such as Mobile Edge Computing. Developing a vRAN solution is the next major virtualization step that will benefit operators and consumers alike.
Currently, we have a centralized architecture with a very efficient core made of storage and compute functionalities, yet still requiring a lot of transport links between the edge and the core. Now a newer architecture is emerging, one that is decentralized, with most traffic staying at the edges—resulting in far less transport. Additionally, there are macro trends that favor this newer decentralized architecture.
Moore’s law is an important trend in technology—its impact on cost will eventually stop the current movement to centralization. We are already seeing signs of this: Significant investment is being made to bring cloud technology near the base station with edge computing.
Centralization enables better management of the infrastructure in the short term, but in the long term, compute and storage costs are decreasing. Fast. The cost of transporting traffic from the edge of the network to the core is also decreasing, but not as quickly.
When comparing the two architectures, the decentralized architecture becomes increasingly beneficial in terms of the costs, storage, and gained efficiencies. So, it looks like virtualization at the edge and in the RAN is here to stay. However, the transition will likely be a slow one.
Virtualization of the RAN is more complex than that of the core. By nature, the RAN is distributed, and centralization of its workload presents challenges. Though the RAN is more expensive than core, RAN virtualization brings a new set of savings and it opens up opportunity for innovation. The benefits will be worth the cost.
While core virtualization meant implementing cloud technology in dozens of data centers or in hundreds of central offices, RAN virtualization will mean implementing technology in thousands of aggregation points and base stations. It also requires deploying general purpose servers alongside RF antenna and backhauling equipment.
On these servers, operators would run traditional base station functions, such as mobility management or baseband software, bringing the benefits of virtualization to the base station. In addition, they will be able to run small versions of packet core software and run new software close to the user when needed. This opens new doors of efficiency and improved security—paving the way to new services and features that will enable the IoT or services such as virtual reality applications.
Despite its many benefits, RAN virtualization will be an evolution rather than a revolution. It will take time to roll out a virtualized RAN, as there are millions of macro cell towers worldwide and hundreds of thousands in the U.S. alone. The rollout of virtualization in each site would be similar to any large-scale rollout of a new technology and will be a multi-year process.
There have been substantial investments in the legacy RAN infrastructure. Though the business case for a greenfield deployment may favor virtualized technology, for communications providers that have legacy infrastructure in place, it is much harder to make a case to rip out and replace that infrastructure with something new.
In addition, with 5G currently in development, timing is not great to start a large scale, national rollout of vRAN hardware. A more practical approach would be to deploy the vRAN hardware when 5G technology gets deployed, which is expected before 2020.
These challenges set the stage for a thoughtful, albeit slow, rollout of virtualization technology in the RAN.
We will, however, see tactical deployment of vRAN technologies in the near future. There will be more proof-of-concept deployments, more trials, many small-scale deployments, and experimentation in targeted areas, as operators look to:
Each of these deployment opportunities will be analyzed closely on a case-by-case basis. Only when the benefits make sense, and the business case is promising, will vRAN be deployed. When utilization is high, but not growing fast, the business case would not make sense. However, there are many instances where it will.
For example, consider a scenario in which an operator wants to begin a small-scale deployment in order to offer LTE service in a rural town of 1,000 inhabitants. In this scenario, a vRAN solution would make much more economic sense than a full-scale solution. With how geographically vast the U.S. is, this is an extremely common scenario in which a vRAN solution could help.
In addition to trials and targeted deployments, operators will be working with industry forums and standards organizations to develop the vRAN infrastructure. This will include making sure that new functions can be decomposed in a way that is consistent with virtual technology. It also includes making sure that management and orchestration systems (e.g. OpenStack) develop the features necessary to support vRAN deployment.
There are strong economic factors in play. The vRAN will evolve. There will be tactical deployments over the next few years, and the technology will be part of the 5G standards. CSPs have time. But they must prepare in order to meet and overcome the operational challenges of vRAN.
Originally published on SDXCentral