Research in the field of Digital Twin technology

Bachelor's thesis, Faculty of informatics and information technologies

Author: David Truhlar, Supervisor: Ing. Matej Petrik

Digital Twin (Core + UERANSIM + Grafana, Loki and custom metrics + LSTM real-time classification with online fine-tunning)

Platform compatibility

This project was tested on:

• macOS Sequoya Version 15.4 with Docker Desktop v4.40.0
• Ubuntu 22.04.3 LTS (Jammy Jellyfish) with Docker v28.0.1
• Windows 11 Home with WSL2 (Ubuntu 22.04.3 LTS) with Docker Desktop v4.40.0

Prerequisites

• git (version 2.39.5)
• docker (version 28.0.1)

Step 1: Clone repository

git clone https://github.com/xtruhlar/5GDigitalTwin.git
cd 5GDigitalTwin/Implementation 

Step 2: Build Docker images

cd ./open5gs/base
docker build -t docker_open5gs .

cd ../../ueransim
docker build -t docker_ueransim .

cd ..

Step 3: Set .env variables

cp .env.example .env
set -a
source .env
set +a

Step 4: Docker compose

docker compose -f deploy-all.yaml up --build -d

Step 5: Add subscribers to MongoDB

docker exec -it mongo mkdir -p /data/backup
docker cp ./open5gs/mongodb_backup/open5gs mongo:/data/backup/open5gs
docker exec -it mongo mongorestore --uri="mongodb://localhost:27017" --db open5gs /data/backup/open5gs

To ensure everything work properly open http://localhost:9999/ in your browser and login using credentials:

Login: admin
Password: 1423

Step 6: Connect UERANSIM gNB to Open5GS Core

docker compose -f nr-gnb.yaml -p gnodeb up -d && docker container attach nr_gnb

Step 7: Connect with registered UERANSIM UE

docker compose -f nr-UEs/nr-ue1.yaml -p ues up --build -d

Then go to Grafana, open http://localhost:3000/ in your browser and login using credentials:

Login: open5gs
Password: open5gs

Open menu on the left, click on Dashboards. Select Current state Dash and you can see the current state of your 5G network.

Example:

Step 8: Run the network simulator (if you want to simulate UC1-UC6)

python network_watcher.py

Code structure

.
├── Implementation
│   ├── data/
│   │   ├── main.ipynb           # Main script for data collection and preprocessing
│   │   ├── datasets/            # Real and simulated CSV datasets
│   │   ├── logs_real_5G/        # Log files from Open5GS components
│   │   ├── Jupyter_Output/      # Jupyter Notebook outputs in HTML format
│   │   ├── current_uc.txt       # Current UC state (UC1-UC6)
│   │   └── running_data.csv     # Exported Prometheus metrics (live)
│   ├── Model/   
│   │   ├── *.ipynb              # Notebooks for model training and evaluation
│   │   ├── trained_models/      # Final .h5 and .keras models
│   │   ├── json/                # JSON files for model architecture
│   │   ├── scaler/  
│   │   │   └── scaler.joblib    # Scaler for data preprocessing
│   │   └── preprocessed_data/   # Numpy arrays for training/testing (X/y)
│   ├── uc1.py ... uc6.py        # Scripts simulating UC1–UC6 behavior
│   ├── network_watcher.py       # Real-time log watcher and classifier
│   ├── open5gs/                 # Configurations for AMF, SMF, UPF, etc.
│   ├── ueransim/                # Init scripts and YAML configs for UEs and gNB
│   ├── promtail/
│   │   └── promtail-config.yaml # Promtail configuration for log shipping
│   ├── loki/
│   │   └── loki-config.yaml     # Loki configuration for log storage
│   └── deploy-all.yaml          # Docker Compose orchestration script
├── docs/                        # Auto-generated Sphinx documentation (HTML)
│   └── index.html, ucX.html, etc.
├── images/
│   └── dashboard.png, model architecture.png
└── README.md

License

Derived from Herle Supreeth's repository, therefore BSD 2-Clause.

Acknowledgements

Special thanks to Ing. Matej Petrík for continuous support and guidance on Digital Twin technology. Ing. Matej Janeba for insights on 5G networks and assistance with collecting real-world traffic data. Herle Supreeth for providing the initial implementation of the dockerized 5G Core.