Table of Contents

1. About The Project
   • Built With
2. Getting Started
   • Prerequisites
   • Installation
3. Usage
4. Information on git branches
5. License
6. Developed by
7. Acknowledgments

About The Project

This project is a POC for our paper "An Intelligent Q-Learning Approach for Energy-Efficient Channel Occupancy in NR-U Cellular Networks for Fair Unlicensed Spectrum Access"

A Simulator developed from scratch in Python to demonstrate various coexistence scenarios for NR-U and WiFi topologies in unlicensed spectrum.

Built With

(back to top)

Getting Started

Instructions to set up the project locally

Prerequisites

• Python version 3.9+

Installation

1. Install Python from official website
2. Clone the repo

   git clone https://github.com/chimms1/NRU-WiFi-Simulator.git

3. Install Python packages (project dependencies)

   pip install numpy pandas matplotlib seaborn tqdm openpyxl

(back to top)

Usage

Branch Info

1. main: has contents of rl-dfs branch
2. rl-dfs: Contains algorithm with 7 states for Q-Learning based Dynamic COT Optimization.
3. dyna-q: Contains algorithm with 21 states for Q-Learning Based Energy-Efficient COT Optimization. Load change can be toggled in ConstantParams and be used with Dyna-Q+.
4. Power-State: (Deprecated, use dyna-q branch) Contains algorithm with 21 states for energy efficient Dynamic Frame Selection based on Q-Learning.
5. CSMA/CA: Development branch used to test implementation of CSMA/CA algorithm present in main file.
6. dev-y: Development branch, used for implementation and testing.

Classes and Files

All the network entities are modeled into various classes

1. Main class Simulator/main-latest-all.py: this is the main file that runs the simulation
2. Service class Simulator/running/ServiceClass.py: contains all the methods used to perform operations such as creating users, calculating resources and many more.
3. Params class Simulator/running/ConstantParams.py: contains the parameters set by the user.
4. Verbose class Simulator/running/Print.py: contains the flags to print specific information and plot graphs.
5. BaseStation class Simulator/entities/BaseStation.py: contains the class definition for NR-U BS and Wi-Fi AP.
6. UserEquipment class Simulator/entities/UserEquipment.py: contains the class definition for NR-U and Wi-Fi User equipment
7. Learning class Simulator/Qlearning/learning.py: contains the class definition for Q-learning (reward function, QTable operation, Actions)

How to execute

1. Set the desired number of users, Number of iterations, Noise, pTx, datarate profile, etc in ConstantParams.py.
2. Set exploration-exploitation iterations accordingly in learning.py
3. Set flags in Print.py to print information.
4. Do additional configurations if required.
5. Run main-latest-all.py

python main-latest-all.py <seed-value>

6. Setting a seed value will help in recreating UE deployments.

NOTE

1. Currently, only a single NR-U BS and WiFi AP can be used.
2. Increasing users to more than 30 may cause a decrease in SINR.
3. Exceptions are not handled in many cases.
4. Internal code may lack documentation in a few places.
5. Complete research work done in this project will be published in 2025-26.

(back to top)

Results Generated from Experiments

The experiment setups and their accumulated data are present in the directory Results-Generation in main branch.

License

Distributed under the AGPL-3.0 License. See LICENSE.txt for more information.

Contributed by

Supervisor: Dr. Vijeth Kotagi

• Yash Deshpande
• Shreyas Joshi
• Ramita Commi
• Gurkirat Singh

Acknowledgments

Resources that we found helpful

• Computer Networks Theory => Computer Networking: a Top Down Approach by Jim Kurose, Keith Ross
• Reinforcement Learning/Q Learning
• DynaQ+
• The readme you are currently reading
• Img Shields

(back to top)