Telecom operators have been on a long journey to implement software-defined networking (SDN) in their data centers and transport networks over the last ten years. Although this hasn’t necessarily disrupted the infrastructure market in the way it was initially envisioned, it has largely succeeded in its goal of centralizing and unifying the control plane of switches and routers, thereby simplifying their management.
However, the mobile market has remained stubbornly opposed to the SDN concept, with control plane algorithms jealously guarded by incumbent vendors inside their base stations (eNodeBs in 4G and gNodeBs in 5G). That is now set to change with the adoption of the Open RAN architecture by most operators worldwide.
Open RAN promises to open the interfaces between RAN elements enabling operators to mix and match different radio and base station components vendors. The base station is disaggregated into a centralized and distributed unit—virtualized software components that run on cloud infrastructure. What’s more, the intelligence inside those components can be decoupled as applications developed by third parties, including the operator themselves. The applications run on top of a so-called RAN Intelligent Controller (RIC) platform, which abstracts the underlying RAN elements, sending the CU and DU instructions to change parameters to achieve an optimization objective or automate an otherwise manual activity.
Such apps are similar to existing Self Organizing Network (SON) use cases in that they can be centralized (running in the so-called non-real-time RIC) and distributed (running in the near RTRIC). But unlike SON, the RIC apps will have access to far more data about the RAN nodes than is currently shared by vendors through the OSS Interoperability Initiative, ingesting information from various network elements in near-real-time (through trace information) as well as non-real-time (through PM counters). Equally, under the influence of the mobile network owners, the apps will have greater freedom to change the configuration and performance of the RAN nodes themselves. Overall, the RICs provide a more unified and standardized method of managing and controlling the RAN than today’s SON.
This doesn’t mean the RAN nodes will become dumb, and the value will migrate to the app layer. Node vendors will still have plenty of scope to differentiate their products and embed intelligence. But they will also have the flexibility to create rApps (centralized or non-real-time) and xApps (decentralized or near real-time) themselves to deliver customer-specific functionality that they do not want to build into their mainstream CU/DU products.
The RICs also promise to enable a thriving ecosystem of independent application developers who have historically been unable to address the RAN market due to the high technological barriers to entry. Those with ML expertise may focus on the longer cycle rApps, while those with a radio heritage may specialize on xApps. But for that ecosystem to take off, the RIC platforms must have friendly APIs and SDKs, which abstract away much of the complexity of the RAN, leaving developers to focus on their application logic.
It is also essential that the RICs communicate with operators’ existing, integrated RANs that do not adhere to the O-RAN Alliance standards. Omdia forecasts that ORAN-compliant vRAN infrastructure will only comprise around 10% of RAN sales in 2025. As such, the vast majority of mobile network infrastructure will not be O-RAN compliant over this period. Like a SON system, the RICs will need to talk with the Element Management System (EMS) of the legacy infrastructure. While this will not afford the same level of control as with open RAN nodes, it should be better than today’s centralized SON (C-SON) solutions. For example, the non-RTRIC will ingest user-level streaming trace data (e.g., Radio Resource Control traces and Minimisation of Drive Test traces, albeit in vendor-specific format), enabling better observability and optimization than SON systems which only ingest cell-level data in 15-minute batches. This user-level visibility could help several monetization use cases that are not possible with SON platforms, such as SD-WAN acceleration and throughput prediction for video adaptive bitrate (ABR) guidance.
What applications will emerge from this new RIC ecosystem? Inevitably many of the early apps will resemble existing C-SON and distributed SON use cases related to automation (e.g., anomaly detection, root cause analysis, site rollout, etc.) and optimization (mobility load balancing, traffic steering, energy saving, etc.). But over time, use cases that enable new monetization opportunities may emerge. Those could relate to network slicing service level agreement (SLA) assurance or geolocation (positioning), for example.
Interestingly, the push towards open RAN and intelligent control is not being led by the vendor community. The service providers are firmly in the driving seat. Our conversations with operators have found an almost universal embrace for the open RAN philosophy and standards.
At a recent webinar, Inside Open RAN Progress: Learnings from the Vodafone RAN Intelligent Controller Trial, Vodafone subject matter expert Carlos Ubeda presented his vision for the RIC as a way to democratize radio resource management in mobile networks. He noted that today’s solutions are proprietary to the RAN node vendors, which leads to vendor lock-in. He sees the RIC as fundamental to improving the cost-effectiveness of the RAN by expanding supply chain diversity and challenging the status quo. Key to this is the open APIs that facilitate the integration of third-party applications.
However, the open RAN movement does not come without challenges. The principal risk that Ubeda identified was ensuring that the pace of specification delivery by O-RAN was maintained and that suppliers commit to implementing these open interfaces.
It remains to be seen whether the introduction of SDN in the RAN in the form of the RIC will be as transformative as many hope. SDN has not radically transformed the transport networks in our view, but we remain hopeful that things will be different in the RAN this time. Done right, the “SDNization” of the RAN will enable operators to:
- Accelerate the pace of innovation in RAN automation, optimization, and monetization
- Easily customize their networks based on their specific preferences and priorities
- Differentiate their network services with new, value-added capabilities
- Diversify the RAN vendor supply chain
- Improve operational efficiency, network performance, and subscriber experience.
This article is sponsored by VMware.