Net AI’s Predictive xApp Running on VMware RIC Cuts RAN Energy Usage by 50%
In common with other responsible industries around the world, the mobile communications industry now has a strong focus on sustainability and energy efficiency. Globally, mobile operators are setting bold targets to make their networks more sustainable by increasing how much of their energy needs come from renewable sources and by expanding their energy efficiency programs. According to the GSMA, operators representing 62% of the industry by revenue and 46% by connections have committed to a science-based target of rapidly decreasing their direct and indirect emissions by 2030. Since the radio access network (RAN) accounts for 80% of an operator’s network energy usage, operators are paying particular attention to how the RAN can be made more energy efficient.
Although improvements in technology make 5G more energy efficient than earlier generations of networks, the density of cells required to provide the coverage and service quality demanded by subscribers places considerable energy demands on the network. Operators can reduce their energy needs by powering off cells at off-peak times, but until now they have been unable to determine whether there are other times they can reduce energy consumption without impacting service quality.
The open and programmable nature of the O-RAN Alliance’s RAN architecture provides a potential solution. By its disaggregated design, the O-RAN architecture fosters greater adaptability, functionality, and energy efficiency compared to legacy RAN designs. The software-centric approach allows a shift in computing resources to large data centres, bringing down power costs compared to local compute operations (See More). O-RAN also introduces the RAN intelligent controller (RIC) as a host platform for a range of RAN management applications, including ones focussed on energy savings. RIC applications can be used to power on and off cells and computing resources assigned to the RAN in response to changing demand.
At the O-RAN Global PlugFest Spring 2023, VMware and Net AI demonstrated an energy savings application that used artificial intelligence models trained with historic and real-time network traffic data to forecast upcoming traffic demands at the level of base stations. The AI models permitted the application to dynamically turn off radio frequency (RF) carriers at the level of base stations when not required, thereby reducing energy consumption in the RAN.
How the Solution Reduces Energy Usage
The solution consisted of Net AI’s Traffic Forecasting and Energy Savings xApp hosted on the near-real-time VMware Distributed RIC platform, with generated RAN data provided by a Net AI E2 Node Simulator and results displayed on a Net AI dashboard. In the simulation, the Net AI application was trained with historical data to predict what future demand would be at the base station level and to anticipate the exact number of carriers required to meet user demands. Accurately predicting traffic enabled the application to use functionality provided by VMware RIC to turn off these carrier or base stations when not required to support subscriber demands, resulting in an average 50% reduction in energy consumption compared to static settings typically used in a legacy RAN, while maintaining over 99% service availability.
The dashboard screen below shows the smart power consumption enabled by the RIC application compared to legacy power consumption (all RF carriers always on) and the optimal theoretically possible power consumption (assuming ideal perfect knowledge of instantaneous traffic demands).
The Innovative Power of a Diverse Ecosystem
VMware is committed to nurturing a varied ecosystem of RIC application developers that can bring a wide range of benefits and features to mobile operators. VMware furnishes a highly scalable, multi-cloud, standards-compliant application hosting platform that fosters a diverse multi-vendor ecosystem, supports vendor interoperability, and empowers operators to tap pioneering solutions. These solutions introduce new use cases — such as automation, optimization, and service customization — that fuel innovation across a radio access network.
According to Lance Uyehara, Director of RIC Product Management at VMware, “The purpose of the RIC is to bring intelligence and programmability to the RAN, and our PlugFest demonstration with Net AI showed a real business benefit of those capabilities. Substantial energy savings in the RAN leads directly to lower energy costs and lower overall OpEx for the host operator.”
Paul Patras, the CEO at Net AI, added that “the PlugFest demonstration showed how our proprietary AI models coupled with VMware RIC enable automated, intelligent decision making in the RAN and facilitate major savings for our customers.”
Find out how VMware and Net AI are helping CSPs take advantage of the O-RAN ecosystem to transform their legacy RAN and enter a new era of efficiency and service offerings:
- The VMware Telco Cloud Platform RAN product page
- The Net AI website
- A detailed solution brief on how the Net AI xApp and VMware RIC work together
For a visualization of how the Net AI xApp and VMware RIC support energy savings, watch this video.
