The wireless industry is abuzz with the promise of open radio access networks (RAN) because of their potential to reduce network costs and make it possible for telcos to roll out new technologies faster and more efficiently.
At its most basic level, open RAN is the decoupling of the RAN functions such as the radio from the baseband and their associated interfaces, which allows wireless operators to use non-proprietary components from of variety of vendors. The O-RAN Alliance is an industry group formed in 2018 by mobile operators, vendors, research and academic institutions to develop specifications for open RAN. Open RAN also incorporates cloud technologies that are used to optimize performance as well as manage the various disaggregated RAN components.
Open RAN is touted as a benefit for wireless operators because it will decrease vendor lock-in, reduce capital and operating expenditures and encourage the deployment of best-in-breed solutions. However, shifting from traditional RAN to open RAN is a big step for many telcos and requires a lot of planning and integration before the technology can be used in commercial wireless networks.
While greenfield operators are embracing open RAN throughout their networks, it’s considered more difficult for existing telcos to migrate away from their traditional RAN equipment to open RAN because of the disruption it creates around traditional telco models.
Nevertheless, the momentum around open RAN is finally growing, and some Tier 1 operators are starting to make big open RAN moves. In late 2023, AT&T announced its $14 billion, multi-year collaboration with Ericsson on open RAN that will include moving 70% of the operator’s wireless network traffic to open platforms by late 2026.
In addition, Canada’s Telus recently revealed that it is working with Samsung on an open RAN 5G network. Telus said commercial deployment will begin in the first half of this year with large-scale rollout expected by mid-year.
Eliminating static service models
However, one downside to deploying open RAN is that the current open RAN architecture limits programmability to certain predefined settings which means that existing, standardized service models, which are how organizations create, test and scale a new offering, are limited.
This means that wireless operators deploying openRAN will have to create new service models if they want to add certain new capabilities, such as interference detection,s to their network. But creating new service models entails working closely with vendors and standard bodies, which can be a lengthy process and slow innovation.
To help solve this problem and make it easier for telcos to get the full potential of open RAN, Microsoft developed a Programmable RAN platform called Project Janus that is based on existing open RAN architecture but allows operators and trusted third-party developers to build dynamic service models without the lengthy review and approval process that is required with the current approach. . Instead, a new dynamic service model can be implemented by an application designer and quickly deployed with assurances that the new feature won’t impact the rest of the network..
For example, because an open RAN requires the integration of different components such as hardware and software from different vendors, performance and network anomalies can occur that disrupt service and affect the performance of the network.
These anomalies are difficult to find and resolve because there are different vendors involved and they all have different ways of troubleshooting problems. In other words, there is no “single throat to choke” because multiple vendors are involved in an open RAN network.
Microsoft’s Project Janus, however, can take care of these anomalies by allowing the service provider or trusted third-party developer to create a dynamic service model and more easily troubleshoot the problem. Using machine learning (ML) and artificial intelligence (AI) to analyze detailed RAN telemetry, the platform can detect network problems. This requires the system to collect raw data and then ML/AI algorithms are used to train the system to find problems and fix them without requiring human involvement. It’s this potential that could result in Project Janus becoming an integral enabler within the newly formed AI-RAN Alliance. The AI-RAN Alliance was announced at the 2024 Mobile World Congress Barcelona conference in February and its goal is to reduce power consumption, enhance mobile network efficiencies and unlock new revenue opportunities for telecom companies through AI.
In addition, when Microsoft’s Programmable RAN platform is combined with an edge data processor it resides at the edge of the network, making it faster and more efficient by delivering sub-millisecond updates and avoiding transmission delays. With response times below 10 milliseconds, this edge processor meets the demands of a real-time RAN Intelligent Controller (RIC), extending the capabilities of today’s near real-time RICs.
Making the RAN energy efficient
Another example of how dynamic service models can benefit open RAN deployments is through energy efficiency. The 5G RAN is the most power-hungry part of the network. Indeed, to meet sustainability goals and reduce costs, some operators choose to entirely power down high frequency antennas during low usage periods, such as overnight, and rely solely on lower frequency coverage during these times.
But power consumption can be more intelligently controlled —even during high-usage periods— using a dynamic service model. For example, a developer can create a dynamic service model to collect the data it needs to create an energy prediction algorithm. This algorithm will then collect data on the number of users on a cell site at a certain time. When there are few users, the algorithm will tell the network to power down the CPU so it is using less energy and then when there are more users, the algorithm will tell the network to power up. Furthermore, with real-time control capabilities, the algorithm can power down a CPU simply because there are no packets that require processing, and then turn it back on when the packet enters the pipeline.
According to Microsoft, an initial prototype of this type of dynamic service model, using a Capgemini 5G RAN and Intel’s FlexRAN reference software, was able to achieve up to 30% energy savings in the network even during busy periods.
New innovations on the horizon
Traditional wireless networks are on the verge of being transformed by the promise of an open RAN. However, current open RAN architectures might be perceived as being somewhat limited in their capabilities. With new innovations such as the programmable edge and development of dynamic service models, telcos will be able to further exploit the capabilities of open RAN and deliver better capabilities and new services to their customers.