This repository contains our work
Graph Neural Networks for Distributed Power Allocation in Wireless Networks: Aggregation Over-the-Air, which is accepted by the TWC.
For any reproduce, further research or development, please kindly cite our paper
@ARTICLE{GNN_aggregation_OTA,
author={Gu, Yifan and She, Changyang and Quan, Zhi and Qiu, Chen and Xu, Xiaodong},
journal={IEEE Transactions on Wireless Communications},
title={Graph Neural Networks for Distributed Power Allocation in Wireless Networks: Aggregation Over-the-Air},
year={2023},
volume={22},
number={11},
pages={7551-7564},
month={Nov.},
}
Instructions:
- Simulation for MPNN, WMMSE and EPA policies can be found in MPNN and WMMSE and EPA.py.
- Simulation for the proposed Air-MPNN can be found in Air-MPNN.py.
- Simulation for the proposed Air-MPRNN can be found in Air-MPRNN.py.
- We give examples for scalability and signaling overhead simulations.
To consider different link densities for testing, change the parameter filed_length in the line test_config.field_length = field_length.
To consider different channel correlation coefficient for testing, change the parameter r in the helper_functions.py.
We thank the works "Graph Neural Networks for Scalable Radio Resource Management: Architecture Design and Theoretical Analysis" and "Spatial Deep Learning for Wireless Scheduling" for their source codes in creating this repository.