Is RIC a Viable Way to Reduce the Energy Footprint of 5G Networks?

At the end of March 2023, total 5G mobile subscriptions reached 1.1 billion and are forecast to reach 1.5 billion with monthly average usage per device increasing to 20 Gigabytes (GB) by the end of 2023. While mobile traffic is growing tremendously, mobile operators do not want power consumption to grow at the same speed. Therefore, reducing power consumption in mobile networks is of paramount importance. In a mobile network, the Radio Access Network (RAN) is considered the most power-hungry component, consuming up to 73%; the core network accounting for 13%; the data center representing 9%; and the rest of operations consuming 5% of the total power consumption according to ABI Research and Next Generation Mobile Networks (NGMN) Alliance analysis. At the cell site, with a traditional RAN, the radio processing, baseband processing, and main controller consume around 50% of the total power, while 40% is consumed separately by radio processing. The power consumption of a 5G base station with Massive Multiple Input, Multiple Output (mMIMO) in a traditional RAN is sub-divided into an Active Antenna Unit (AAU) and a Baseband Unit (BBU), with AAU accounting for around 90% of the total power consumption. Considering the above proportion, it is clear that reducing the power consumption in RANs, especially at the radio site, will significantly improve the overall energy efficiency of mobile networks.

Open RAN is now positioned as a catalyst for energy efficiency. The Open RAN architecture offers several energy-efficiency features, including intelligent control via a RAN Intelligent Controller (RIC), fronthaul functional split, and use of Common-of-the-Shelf (COTS) hardware. The RIC is one of the features supported by Open RAN that provides optimized intelligent control of the RAN by aggregating several Key Performance Measurements (KPMs) related to the status of network infrastructure, such as radio resource utilization, number of users, throughput, load, power usage, etc. Moreover, with the advent of Open the RAN fronthaul functional split, significant amount of energy reduction can be achieved by using split option 7-2x by offloading parts of PHY (L1) processing to the Distributed Unit (DU). It should be noted that legacy vendors also offer the same features, but through proprietary and highly customized features, which often lead to vendor lock-in and lack of interoperability.

Open RAN also offers improved technology for better sustainability. The radio component consumes the highest amount of power in a RAN and the consumption is divided into Power Amplifiers (PAs), small signal, Digital Intermediate Frequency (DIF), and power supply. It is noted that the power consumption of a RAN varies with the traffic load that significantly impacts the overall power consumption ratio of each component in mobile networks. The PA accounts for the highest proportion of power and consumes around 60% under the full load considerations. Therefore, reducing the power consumption at the equipment level plays a significant role in achieving energy efficiency in mobile networks. One effective solution is the use of highly-efficient semiconductor materials, such as Gallium Nitrate (GaN), to maximize the power density, which may result in increased energy efficiency up to 40%. For example, GaN has been prominently used by Huawei, ZTE, and other vendors, providing significant reduction in power consumption of their Remote Radio Units (RRUs)/AAUs. Another option is to leverage the concept of multi-band radio, with multiple bands served by a single PA. Another effective approach is “sleep mode” to eliminate underutilization of spectrum resources to meet peak hours demands by shutting down base station components to save energy. There are different levels available for sleep mode, including symbol shutdown, mMIMO muting/carrier shutdown, and traffic shaping.

There are several promising features of Open RAN identified by the O-RAN Alliance that have the potential to achieve targeted KPIs of next-generation mobile networks. The most important feature of Open RAN is the near-Real-Time (RT) RIC that has the ability to control and optimize RAN elements (Centralized Unit (CU), DU, and RU) via fine-grained data collection and intelligent decisions over the southbound (E2) interface. One of the key components of near-RT RIC is the xApps, which is a plug-and-play unit designed to run on the near-RT RIC to define custom logic for the RAN. The RIC can process intensive data by leveraging Artificial Intelligence (AI), Machine Learning (ML) and Federated Learning (FL) algorithms to provide policy-based guidance for the RAN to optimize radio resource management, RAN slicing, traffic steering, handover process, interference management, beamforming optimization, scheduling, and, last but not least, power management to achieve higher energy efficiency in the RAN. The optimization of all these features will significantly increase both the spectral and energy efficiency in the RAN.

However, the RIC is currently under development, and it might take a couple of years before its commercial deployment and integration with RAN components. Several vendors have contributed to O-RAN Alliance Work Group 2 and Work Group 3 to define the architecture, interfaces, and use cases for non-RT RIC and near-RT RIC. Moreover, the requirements and subscription management of xApps has already been defined and vendors have started to develop their own xApps to support their RIC platforms. There is an ongoing competition between vendors in terms of RIC development. Nokia recently announced its MantaRay RIC to support new use cases by implementing Open RAN standardized RIC capabilities. For example, Nokia has designed its Traffic Steering xApp based on AI/ML-based algorithms to optimize traffic distribution in a RAN. Ericsson has designed its Intelligent Automation Platform that includes a non-RT RIC for running radio network applications (rApps). Earlier this year, Ericsson developed two rApps, including Ericsson’s RAN Energy Control and RAN Energy Cockpit to optimize energy efficiency in the RAN. As an alternative to the RIC, Huawei has developed its own Intelligent RAN that was initially introduced at MWC 2022 and more features were announced at MWC 2023.

Currently, the industry is finding ways to deploy RICs in commercial deployments, while Huawei has claimed that its Intelligent RAN solution has already been deployed in some regions across Europe and Asia. Depending on such developments, one could question how much time it will take to commercialize RIC use cases, when will we see its full integration with RAN components, and when will the full potential of a RIC be used to enhance energy efficiency in a RAN? Based on recent developments in this domain, including standardization of RIC use cases, rApps, xApps, and interfaces, ABI Research forecasts that RICs will be commercially available by 2H 2024 because there is an increased amount of interest and competition among vendors to become first to the market, which may further accelerate this development to happen sooner than later. Moreover, RICs must be tested before they will be integrated with RANs to support new applications and services for operators in brownfield networks. The testing phase for RICs has recently started and several operators, in partnership with vendors, have conducted field trials for non-RT RIC and near-RT RIC. For example, BT conducted a non-RT RIC testing in collaboration with Nokia in the United Kingdom. Vodafone announced the testing of its 5G site using RIC in collaboration with multiple vendors, including Cohere Technologies, Intel, Telecom Infra Project (TIP), VMware, and Capgemini. It is expected that the testing of RICs will continue to happen by the end of this year before their commercial deployment later next year.