ChaogeCanFly / Drone-Blind-as-a-bat----AngleRealizability

Code for ICASSP 2020 paper on angle realizability

Realizability of planar point embeddings

This repository contains the code to reproduce all results of the 2020 ICASSP paper named Realizability of planar point embeddings from angle measurements.

@inproceedings{Duembgen2020,
  author={Dümbgen Frederike and El Helou Majed and Scholefield Adam}, 
  title={Realizability of planar point embeddings from angle measurements}, 
  booktitle={IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  year={2020}, 
  pages={xxxx--xxxx}
}

Authors

• Frederike Dümbgen
• Majed El Helou
• Adam Scholefield

Installation

This code has been tested with python3.5.7.

All requirements are available through pip and can be installed using

pip install -r requirements.txt

Use code

This code base was developed for theoretical analysis and we do not guarantee efficiency or user-friendliness. However, if the reader is interested in further development, the best starting points for learning how to use the code are the notebooks.

Reproduce figures

The Figures in the paper can be reproduced using the two below notebooks.

• Analysis.ipynb: Figure 2.
• Realizability.ipynb: Figure 3.
• Angles_vs_Distances.ipynb: Figure 4.

Figures 3 and 4 use pre-computed results. To generate new results, you can run

• simulation_discrepancy.py: apply increasing number of constraints for denoising (used for Figure 3).
• simulation_angles_distances.py: angle vs. distance-based localization (used for Figure 4).

To make your life easier, you can generate all results from the paper in one line by running

./generate_results.sh

or, if you just want to have a quick set of results, run

./generate_results_quick.sh