mobile-network-datasets-ns3

This project contains scripts to generate datasets from ns-3 4G and 5G mobile networks. It comes packaged with the base ns-3 module along with additional third party ns-3 modules that are required to generate diverse datasets.

It also contains scripts to parse the data, do sanity checks on them and visualize it in various ways.

Finally it also contains the machine learning code repository that has the code for the results in the paper "Self-supervised Pretraining for User Performance Prediction under Scarce Data Conditions"

Install instructions

This is a full ns-3 installation along with the additional modules and changes to the base code.

Step 1: Clone the repository

git clone https://github.com/akhila-s-rao/mobile-network-datasets-ns3.git

Step 2: Go into the ns-3 code folder

cd mobile-network-datasets-ns3/ns-3-dev

Step 3: Configure the ns-3 code

./waf configure --build-profile=optimized --enable-examples

This creates an optimized build which disables debug mode and improves simulation speed. To enable debug mode which allows you to print logs from the NS_LOG_COMPONENT

./waf configure --enable-examples

Several components will not be configured and be shown in red. This is normal.

Step 4: Build the code

./waf build

Step 5: Create a folder to store your logs

mkdir logs

Step 6: Tune parameter knobs, modify the simulation scenario setup and run a simulation campaign

cd mobile-network-datasets-ns3/ns-3-dev/src/nr/examples/dataset_gen_scripts edit run_script.sh as required bash run_script.sh

Curated datasets generated from these scripts can be found at (google drive link)

https://drive.google.com/drive/folders/1-OQolvhK1mpFNFVLF4XTC_c3_EjLPwz-?usp=sharing

Acknowledgement

This work was funded by the H2020 AI@EDGE project (grant. no. 101015922)