This respository contains power consumption measurements of open-source 5G Core Networks deployments.
- A. Bellin, M. Centenaro and F. Granelli, "A Preliminary Study on the Power Consumption of Virtualized Edge 5G Core Networks," 2023 IEEE 9th International Conference on Network Softwarization (NetSoft), Madrid, Spain, 2023, pp. 420-425, doi: 10.1109/NetSoft57336.2023.10175489.
- Extended version [WIP]
Please cite our published papers if you intend on using the data in this repository
The testbed used to gather the data comprises 5 Intel NUC units, the 3 used for the 5GC deployments have the following specifications:
- i5-7260U processor @2.20GHz
- 8GB of RAM
- 240GB SSD
- 1 Gigabit Ethernet and Dual Band Wireless connectivity
- Ubuntu 20.04 Desktop OS
We also used the follwoing software:
- Open5GS
- Free5GC
- UERANSIM
- Redis
- Scaphandre
- Iperf3
The power consumption data is gathered using both hardware-based (Meross MSS310 smart-plugs) and software-based (Scaphandre) power meters.
The data is subdivided between Scaphandre measurements and smart-plug measurements.
Each category is then subdivided between data gathered while using Open5GS and Free5GC.
Open5GS has been tested with three alternative virtualization options:
- Virtual Machine (VM)
- Bare Metal (BM)
- Docker Containers (CO)
Free5GC has been tested only on bare metal.
All data is stored in JSON file format. Each file contains the data for a specific data-plane throughput level specified in the filename.
In the Scaphandre collection we saved the output of the JSON exporter without further processing.
For the smart-plug collection, other than power data, the files also include thorughput and cpu load metrics collected using the psutil Python library. We also include the timestamp for each data sample.