5G Traffic Capture Dataset: User Equipment Classification in the 5G Core
Authors/Creators
-
1.
Universidade Federal de Juiz de Fora
- 2. Federal University of Juiz de Fora
Description
A 5G dataset containing network traffic capture files created on a simulated environment and saved in the PCAP format
These datasets were used in the masters dissertation titled "User Equipment Classification in the 5G Core" according to the description in the Table of contents below
Technical info
The network packets of the inference dataset were captured using tcpdump on Ubuntu Linux and saved using the PCAP file format (which is more detailed than CSV)
The virtual environment used free5GC and UERANSIM to simulate the 5G network
Table of contents
Dataset file description
| File name | Description |
| training_data.zip | Data from other public datasets used to train the ML models |
| inference_data.zip | Data collected and used to run the model inference step |
| File name | For training or inference? |
5G traffic type | Description |
| Youtube_cellular.pcap | Training | eMMB | Video reproduction traffic collected from the "5G Traffic Datasets" public dataset |
| naver5g3-10M.pcap | Training | URLLC | Video live stream traffic collected from the "5G Traffic Datasets" public dataset |
| training_data_mmtc.zip | Training | mMTC | UDP traffic collected from the "5G Campus Networks: Measurement Traces" public dataset |
| youtube-1M-1080p.pcap | Inference | eMMB | Captured while virtual UE reproduced the "5G eMBB traffic" Youtube playlist |
| naver-tv-1M.pcap | Inference | URLLC | Captured while virtual UE reproduced the NaverTV's YTN "Korea 24-hour news channel" live stream |
| udp-100pps.pcap | Inference | mMTC | Captured while virtual UE used our traffic generator/Netcat to send 100 UDP packets per second |
| udp-nc-traffic-1k.pcap | Inference | mMTC | Captured while virtual UE used our traffic generator/Netcat to mimic an IoT-like UDP communication pattern (with variable intervals) |
Notes
The data was split in two ZIP files (training_data.zip and inference_data.zip). If using them with our source code, they should be extracted to the same folder called "pcap". For more details on the expected folder structure, see the directory structure files in this other Zenodo record
Notes
The training data was obtained from the IEEE Dataport "5G Traffic Datasets" and "5G Campus Networks: Measurement Traces" public datasets. The files' contents were not changed.
Files
inference_data.zip
Additional details
Related works
- Has part
- Dataset: 10.21227/ewhk-n061 (DOI)
- Dataset: 10.21227/xe3c-e968 (DOI)
- Is supplement to
- Annotation collection: 10.5281/zenodo.15473395 (DOI)
Funding
- Coordenação de Aperfeicoamento de Pessoal de Nível Superior
- National Education and Research Network
- Fundação de Amparo à Pesquisa do Estado de Minas Gerais
Dates
- Created
-
2025-02-13Inference dataset was created on this date
Software
- Repository URL
- https://github.com/netlabufjf/nwdaf_ml
- Programming language
- Shell , Python
- Development Status
- Active