This repository contains the Python and MATLAB codes for the paper Real-time indoor localization for smartphones using tensor-generative adversarial nets. IEEE Transactions on Neural Networks and Learning Systems, 2020, by X.-Y. Liu and X. Wang. [https://ieeexplore.ieee.org/document/9159909]
- A 20 m × 80 m floor (located in a research institute building).
There are in total 30 randomly deployed APs; fingerprints are sampled at grid size: 0.3m x 0.3m.
On the fourthe floor, it can receive 21 APs, and an orginal 63 x 268 x 21 tensor was collected.
We preprocess it into a fingerprint tensor: 64 x 256 x 21 (copied the 63th row to a 64th row; deleted the 257th to 268th columns).
-
A real-dataset: WiFidata_real.mat collected for our Smartphone App.
size: 6 x 16 x 14; and sample grid size 0.1m x 0.1m.
-
We also include a synthetic fingerprint dataset for testing tensor completion algorithms. Size: 476 x 598 x 15, used in a previous project: Adaptive sampling of RF fingerprints for fine-grained indoor localization. IEEE Transactions on Mobile Computing, 2016. By X.-Y. Liu, S. Aeron, V. Aggarwal, X. Wang, M.-Y. Wu.
It was generated by a software 3D Indoor Standard Ray Tracing (software) A. Kaya, L. Greenstein, and W. Trappe W, “Characterizing indoor wireless channels via ray tracing combined with stochastic modeling,” IEEE Transactions on Wireless Communications, 8(8): 4165-4175, 2009.
We used MATLAB for kNN and Direct neural networks
-
kNN_performance implements the kNN algorithm
Test_kNN_Plot.m tested the localization performance and dras CDF curve of localization error.
-
Direct neural networks (it split the data into training set and testing set, we also include the data in the MATLAB folder).
It is a regressor, and we used the matlab deep learning toolbox with GUI, at link: https://www.mathworks.com/products/deep-learning.html
The file RF_regression_net.m is saved from this toolbox, for a reference.
-
For tensor compleiton, we worked on MATLAB and put it into a saparate folder.
tensorflow 1.8.0 python 3.5.3 numpy 1.14.2 pillow 5.0.0 pickle 0.7.4
A code to implement the original GAN.py with a generator and a discriminator. Testing on the mnist data.
run GAN.py to train a generative model which is used to generate image data from a random distribution
test data gen: gen_test_data.py
show results: draw_pic.py
A code for a TGAN.py with a regressor and a discriminator.
As last, I tested few python codes for kNN and Direct Neural Networks (not tuned).