The 5G-ECOnet project investigates the combination of the necessary energy-efficient operation of Open RAN-based 5G campus networks while maintaining the required quality of service for applications in industrial environments. This includes the applicability of energy saving measures, their adaptation as well as their conformance or integrability in standards. A central question is also to what extent network elements and functions can be switched off or reduced in their resource requirements without affecting the quality of service and the industrial application. In addition, examples are presented and evaluated to what extent renewable energies can be usefully applied in the field of communication infrastructure, especially in industrial campus networks.
In the 5G-ECOnet project, the TU Ilmenau is investigating the influence of the use of AI processes on the energy consumption of 5G networks. Thus, using the example of network management functions for campus networks, the influence of the realization of AI processes on energy consumption is to be investigated. For this purpose, machine learning (ML) methods implemented as rApp in the Open RAN ecosystem will be realized and compared with respect to energy consumption and performance. The focus of the investigations will be on functions for optimizing network coverage, network capacity and improving interference in the network. In particular, federated learning methods will be compared with centralized learning methods and the influence of design and operating parameters of ML methods on energy consumption and quality of service of the network will be investigated. In addition to theoretical investigations, the impact of exemplary improvements on 5G campus networks will be demonstrated in practice.
Partners in the project are Fraunhofer-Institut für Integrierte Schaltungen IIS, the AiVader GmbH, exceeding solutions and Keysight Technologies.