JiahongChen / WSN-SAE

This repo provides source code for optimizing sensor sampling locations in wireless sensor networks using spatiotemporal autoencoder.

WSN Sampling Optimization using Spatiotemporal Autoencoder

This is the Python+TensorFlow code to reproduce results for paper 'WSN Sampling Optimization for Signal Reconstruction using Spatiotemporal Autoencoder'.

Due to GitHub file size limitations, datasets are not upload to this repo, you can:

1. Download raw data from NOAA.
2. Request preprocessed data by sending email to me at [e-mail].

Requirements

• Platform : Linux
• Computing Environment:
  • CUDA 10.1
  • TensorFlow 1.14.0
• Packages: pandas, numpy, scipy, argparse.
• Hardware (optional) : Nvidia GPU (SST requires around 7GB of GPU memory)

Getting Started

1. Computing environment set up can be refered to this repo.
2. Download data and place it at './Data' folder.
3. Run the code by

bash batchrun.sh

Citation

Please cite our paper if you use our code for your work.

@article{chen2019optimization,
  author={J. {Chen} and T. {Li} and J. {Wang} and C. W. d. {Silva}},
  journal={IEEE Sensors Journal}, 
  title={WSN Sampling Optimization for Signal Reconstruction Using Spatiotemporal Autoencoder}, 
  year={2020},
  volume={20},
  number={23},
  pages={14290-14301},
  doi={10.1109/JSEN.2020.3007369}}

To do

• Code for visualization
• Code for optimizing WSN sampling strategy