This repository contains the entire code for our TWC work "Graph Embedding based Wireless Link Scheduling with Few Training Samples", available at https://ieeexplore.ieee.org/document/9285223.

For any reproduce, further research or development, please kindly cite our TWC Journal paper:

M. Lee, G. Yu, and G. Y. Li, "Graph embedding based wireless link scheduling with few training samples," IEEE Trans. Wireless Commun., vol. 20, no. 4, pp. 2282-2294, Apr. 2021.

How to use?

Requirements

The following versions have been tested. But newer versions should also be fine.

• rdkit : [Q3 2017 Release](https://github.com/rdkit/rdkit/releases/tag/Release_2017_09_1, Release_2017_09_2)
• boost : Boost 1.61.0, 1.65.1

Setup

Go to "s2v_lib".

Build the c++ backend of s2v_lib and you are all set.

cd s2v_lib
make -j4  

Note: We utilize existing open-source code for the structure2vec architecture (https://github.com/Hanjun-Dai/pytorch_structure2vec/tree/master/s2v_lib) and add the batch nomarlization function in it.

Data Prepareation

Go to "FPLinQ".

Run "generate_main.m". The output data is saved in "/mat".

Copy the output data into "D2D_qua/mat".

Run

Go to "D2D_qua".

Run the main program.

./run.sh

You can also use the following command.

python main.py