Among the most celebrated principles in the Open RAN paradigm, softwarization and disaggregation have taken center stage in discussions on the commercialization and deployment of O-RAN systems. A less frequently discussed principle that these two elements enable is programmability — the ability to dynamically and algorithmically adapt the RAN’s configuration and behavior to optimize performance for specific conditions and use cases. With programmability and open interfaces comes closed-loop control, allowing for the use of telemetry and data from the RAN itself (and potentially other sources) to automatically assess the RAN’s state, match it with the optimal action or configuration and apply it within the programmable RAN system.
Programmability and closed-loop control enable energy optimization within the RAN, support for diverse traffic profiles with conflicting requirements through dynamic slicing, and dynamic load balancing of users across the network, among other benefits. These capabilities are achieved through a combination of artificial intelligence (AI) and machine learning (ML) techniques that utilize data streams and telemetry exposed by the RAN for control, classification and prediction. This architecture also facilitates the integration of different viewpoints on the RAN, ranging from highly granular, device-specific views at individual base stations to more general, centralized views that aggregate data from tens of base stations and hundreds of users.
Programmability and closed-loop control have thus the potential to redefine how the RAN is managed and optimized. These principles are embodied in the O-RAN RAN Intelligent Controllers (RICs), which provide control and analytics at non- and near-real-time scales. The non-real-time (non-RT) RIC handles broader orchestration and policy-definition tasks, operating on loops of one second or above. The near-real-time (near-RT) RIC manages control loops at intervals between 10 milliseconds and 1 second, directly influencing network performance through radio resource management in the RAN. The non-RT RIC manages scalable control policies across thousands of devices, while the near-RT RIC ensures fast, localized responses to network conditions.
There is a disconnect, however, between the importance and value that the RICs, including the Near-RT RIC, bring to the O-RAN architecture and their commercial adoption. ATIS recently released a Minimum Viable Profile (MVP) document for the North America region which includes the Near-RT RIC as an optional component. The availability of E2 and O1 interface implementations on commercial RAN stacks is scarce, with open-source frameworks representing the most advanced solutions in the space. Overall, this points to the challenges that the RIC ecosystem and the overall Open RAN community still needs to address to make intelligent and programmable wireless networks a reality:
Complete specifications, including testing and interoperability profiles.
These are required at various levels and across different interfaces, including those between the RICs and the RAN, as well as between the application logic and the RAN (e.g., testing E2 service models across xApps and the RAN). As functional specifications rapidly evolve — introducing innovations every six months — there is a growing need for automated and continuous testing, integration and validation throughout the end-to-end ecosystem, considering both the RAN and the applications running on the RICs. To address this, O-RAN is developing test specifications for RIC interfaces, and O-RAN PlugFests are intensifying efforts to test RIC-related interfaces. Additionally, testing must also encompass the energy efficiency of these systems. From the continuous integration point of view, we recently released a framework that streamlines the support of open-source stacks such as OpenAirInterface and srsRAN for xApps on the O-RAN Software Community Near-RT RIC.
Manage the network complexity. The RICs and associated AI-driven Open RAN software components introduce additional complexity in terms of configuration, versioning and resource and infrastructure management. Addressing this complexity through intelligent automation has been a key focus of our research, resulting in the development of an intelligently orchestrated operating system for the deployment of end-to-end, full-stack, fully managed O-RAN systems. This solution is now being commercialized through a Northeastern spinoff, zTouch Networks.
Design of efficient and effective AI/ML for the RICs. More research is needed on AI/ML for the RIC to ensure the design of algorithms that are both efficient and effective across diverse network conditions. AI/ML-based control solutions, once trained, must also undergo rigorous validation and testing in controlled environments to minimize the risk of disrupting production networks. However, these testing environments must also be realistic enough to produce meaningful results, factoring in user load, traffic patterns and RF characteristics reflective of real-world deployments where these models will eventually be implemented. By improving both the design and testing of AI/ML models, we can better ensure their reliability and performance in dynamic network environments.
Data and role of digital twins. Developing robust and scalable AI/ML solutions that generalize effectively across diverse real-world deployment scenarios requires leveraging comprehensive datasets of RAN telemetry, operational data and performance metrics. While network operators have the capability to collect such datasets, privacy and security concerns often render them impractical for use in research and development. Wireless digital twins, such as Colosseum, incorporating embedded O-RAN environments, provide a viable alternative to address these challenges, enabling the creation of accurate and secure testbeds for AI/ML model development and validation.
Extensions to the real-time domain. Today’s RICs control domain concerns time scales above 10 ms, or one 5G NR frame. In addition, the optimization only interacts with the control plane, without access or influence on user plane data units. Real-time control and interaction with the data plane, however, are key enablers of dynamic control in a variety of use cases, from spectrum sensing and sharing to fronthaul optimization and integrated sensing and communications. In this context, the O-RAN ALLIANCE nGRG has taken on a research item on dApps, a real-time extension of the RIC architecture that will provide plug-and-play programmable loops within the RAN itself. This research item has, so far, received contributions from Northeastern University, NVIDIA, Mavenir, Qualcomm, MITRE and reviews from Ericsson, Samsung, Verizon, Jio and Keysight.
Addressing these points is key to the development and adoption of programmable, intelligent closed-loop control in mobile systems, as well as to unlocking key performance gains through agile and bespoke network configurations.
