posted 12th May 2025
This EUCNC 2025 exhibition booth provides an in-depth presentation of five selected SNS projects, notably BeGREEN, 6G-TWIN, 6GREEN, EXIGENCE, and 6G-SENSES, that are explicitly targeting concrete ecological improvements of 6G, going from energy efficiency over carbon footprint reduction to net energy consumption limitation. Hence, each of the participating projects addresses specific challenges, yet the presented solutions are complementary in that they exhibit strong potential for synergetic integration on a system level. Together, these projects therefore could contribute concrete and quantifiable improvements towards an overall more sustainable 6G. Common to all involved projects is the strong shared belief that the technology improvement will have the strongest network effect, therefore positively influencing other aspects, such as economic and societal sustainability, in a longer run.
In this sense, the presented demos cover different aspects, ranging from RAN energy efficiency enhancement by introducing RAN (and network) monitoring and optimisation, over new architectures and enablers for improving energy efficiency of the future network, i.e., Cell-Free MIMO and Integrated Communication and Sensing, a proposed Observability Framework for monitoring energy consumption of the physical and virtual components of the network, to green, live ecolabels for future 6G services, empowering service users and incentivising them to an ecologically-aware usage.
BeGREEN Intelligence Plane Demo
This demo will showcase the baseline implementation of the Intelligence Plane, concretely the AI Engine and the non-RT, integrated with the dRAX and Emulated RAN via VIAVI TeraVM AI RAN Scenario Generator (RSG). The AI Engine will host two ML models, a cell load predictor, and a cell energy predictor, already trained with offline data. Then, “AI Engine Assist rApps/xApps” will be deployed in the RICs to assist ML model inference, exposed to a control rApp. Finally, a RAN control action based on the ML model output will be sent to the dRAX and the Emulated RAN, so action could be seen in real time. Visitors will see the results on a specific Grafana dashboard to showcase the different outputs of the demo. Here, the A1 policies created by the rApp and their implication on the Project KPI such Energy Score and Power consumption reduction will be presented.
6G-TWIN Demo
6G-TWIN is advancing the use of Network Digital Twins (NDTs) to optimize energy management in AI-driven 6G systems. Building on the BeGREEN Intelligent Plane demo (see above item 1), this demonstration will showcase how NDTs can enhance energy efficiency in O-RAN networks. It will leverage the BeGREEN setup, integrating Accelleran’s Near-RT RIC with xApps, i2CAT’s non-RT RIC with rApps, and the BeGREEN intelligence plane, while introducing NDTs as a key enabler for future AI-native 6G architectures. Using the VIAVI AI RSG, the demo will emulate real-world scenarios to illustrate ongoing work in energy-efficient network operations. Visitors will see an integrated Emulated RAN with telemetry capabilities to extract information to create an NDT in a dedicated Dashboard. The figure describes how the 6G-TWIN Architecture is being mapped to the Energy Saving Use Case (UC2), and how the Telemetry layer is implemented to provide telemetry for the NDT creation.
Visitors will see aggregated extracted telemetry needed to create an NDT in a representation dashboard. The demo is designed to be visually engaging and interactive, ensuring its accessibility to a broad audience. The use of the VIAVI AI RAN Scenario Generator provides an intuitive interface that simplifies complex concepts like energy-saving strategies and network intelligence. Clear and concise explanations, paired with a focus on sustainability and innovation, ensure the exhibit resonates with both technical and non-technical visitors.
BeGREEN / 6G-SENSES Integrated Communication and Sensing (ISAC) Demo
This demo will consist of a multi-technology sensing demonstration (Sub-6, and 5GNR) where an O-RAN-based system will be tailored to integrate an ISAC service model implementation, sensing Apps, implemented as xApps, and additional HW (3GPP and non-3GPP). The setup will comprise: 1) a Sub-6 ISAC platform (developed in the context of BeGREEN) that is built up of Ettus N321 RUs that will act as radar-based sensing nodes, and 2) a server hosting the DU, CU and RICs and an O-RU. The received uplink SRS signal will be received at the RU to perform sensing. These systems will make available the sensing signals to be post-processed in the form of a heat map, which can be displayed with Grafana dashboards.
People moving around the Exhibitor will be detected by the Sub-6 sensing system, and the integration of both sensing sources (3GPP and non-3GPP) will be showcased. The diagram that depicts the ISAC demo.
In addition to this, BeGREEN will provide a demonstration on sensing-assisted communications using RIS at Sub-6 frequencies. The idea of this demonstration is to use a RIS board to extend the coverage of an RU using the reflective path of an RIS to communicate through an obstacle, which allows turning off one of the RUs (supposed to cover that space) offloading the users in a specific area.
6G-SENSES CF-mMIMO Demo
The intention of the demo is to show the principle of a cell-free network architecture with different parameters.
The demonstration itself incorporates a simulation of a (small-scale) cell-free network architecture. The algorithms included are assignment of UEs to APs, Beamforming/Precoding, and Power Allocation, which are performed in respective order and not jointly. There is no scheduling across Resource Blocks or similar involved.
The demo provides a UI-window showing UEs/APs and their computed connectivities in a 2-dimensional view to illustrate the cell-free approach dynamically to the audience. For interaction, the user can change different network parameters, such as the number and location of UEs/APs and the size of the network area. Also, switching to different optimization algorithms is possible. After changes to the network settings, the UI of the demo then shows the results of the newly computed optimization procedures.
6GREEN Demo
This demo will showcase the 6Green Observability Framework, able to combine information coming from both the physical infrastructure and the 5G network to infer the consumption ascribable to the virtualized components (VM or container) running on top of it. The visitors will see in real-time how, upon transmission of mobile traffic, the Observability Framework allows to break down the HW-level consumption and map it onto several 5G network functions. Moreover, the visitors will also see how the CO2eq emissions change by considering the energy grids of several countries and energy sources.
As the whole system, from the transmitted traffic to the shown metrics is configurable, visitors will be able to ask for specific metrics to be shown such as number of UEs, PDU sessions, transmitted traffic, etc.
As shown in the diagram above, the access network will be simulated by using LoadCore, while the 5G core will be one of the main open-source releases. Finally, results collected in real-time will be visualized by using Grafana:
EXIGENCE Demo
This demo shows how we collect and visualize live energy consumption measurements for any service instance within a future 6G system. The service can be based on a single server/node, or it can be distributed, involving several nodes and communications among them. Concretely, the live demo shows:
• A supported service is deployed within the system and usage starts. (Currently supported: single node task execution; video streaming to one or multiple clients; distributed AI computation).
• Service usage can be energy monitored if required. We show how we deploy energy metering probes wherever relevant service elements reside.
• We then get energy consumption information for each such monitored service. We show how energy consumption increases when the usage starts, intensifies, shrinks, etc.
• We show how energy consumption changes with service parameters (e.g. computational intensity, video resolution, etc.).
• We show that we can aggregate this information by service type, by user type, etc.
• We show tight integration with physical power meters that send data via MQTT to an energy collection server. This helps us validate software-based energy measurements.
The figure above shows first phase. Based on the assumption that in 6G in-network-computing can be used to provide beyond-connectivity services from a mobile system, we first deploy such service and then show energy consumption information collection for the latter or even for a given instance of it. EXIGENCE displays that information to the resp. service user. Here, we demonstrate the feasibility of measuring energy consumption using software and hardware meters and the importance of showing this to the service user.
In a second demo scenario, we remotely log in into a real remote 5G testbed, from project partner Internet Institute, in Ljubljana, Slovenia. We deploy the same services remotely yet still display the energy consumption of the whole remote 5G testbed including core, RAN, and the application at the booth.
Measurements using software and hardware meters are used to present the user with the energy consumption of the used services as shown in the following graphs.
In summary, Exigence demo proves the feasibility of energy monitoring of individual logical allocations in an infrastructure (e.g. one service instance). The secondary insight of the demo is that energy consumption estimation based on crude models, e.g. on data volumes, is inaccurate. Finally, the demo shows service-level optimization potential.
The demo is shown over a local setup with WiFi only, seamlessly extended by a remote 5G setup in a partner test lab location at Internet Institute in Ljubljana, Slovenia. (Local 5G installation is difficult due to unclear frequency license situation).
BeGREEN DU Hardware Acceleration
Sphere Decoder (SD) is a near Maximum Likelihood (ML) MIMO demodulator. Its advantage is that it can utilize spatial diversity much more efficiently than linear receivers and improve the MIMO decoding performance significantly. The way an ML demodulator works is that it checks a Euclidean distance metric from the received signal to all possible combinations of the transmitted symbols and finds the minimal distance. The number of metrics that needs to be calculated is significantly high – for example for 256QAM and 4 layers for implementing a full ML demodulator we would need to go over 256^4 which is ~4 billion metrics. The SD does go over distance metrics like the ML demodulator but reduces their number significantly by containing them in a multi-dimensional sphere. Still, their number is large, which requires a very large number of operations which can be done in parallel. In the demo we will show how running the SD on a GPU with its parallel computing architecture\is more power efficient than running it on a CPU. The platforms that will be compared are the Arm Neoverse N2 running on the Marvell Octeon 10 CRB (CPU) and the 2048-core NVIDIA Ampere GPU with 64 Tensor Cores running on the Jetson AGX Orin 64GB (GPU).
BeGREEN RU Power Consumption Improvement by PA Blanking
This demo describes the purpose of the Power Amplifier Blanking Module in the B5G Radio Unit (RU), as well as the module architecture, its operating mode and the RU power consumption savings that can be achieved out of this module.
