Copyright (c) 2017, Tampere University of Technology (TUT)
This work is licensed under the MIT license / X11 license.
This documentation is licensed under CC0 license.


This folder contains the following Matlab/Octave files

 run_pos_est                                main file to run 
 fnc_read_CSVdata_inMatlab.m                Load the database
 fnc_coverage_distrib_based                 pointwise coverage area algorithm
 fnc_coverage_pointwise		 	    coverage area defined with distribution algorithm
 fnc_eval_error				    evaluates error in a set of 3D position estimates
 fnc_get_data_properties		    picks defined outputs of rss_stats
 fnc_get_floor_heights			    finds the floor height values			
 fnc_get_knearest			    finds the k rows in an matrix data
 fnc_rss_stats				    compute positioning error statistics

There are two implementations: 1) pointwise coverage area, and 2) coverage area defined with distribution.

The obtained accuracies are:

1) pointwise coverage area (fnc_coverage_pointwise)
     mean_error_3D: 10.0397
     mean_error_2D: 9.4434
    floor_det_prob: 0.8664

2) coverage area defined with distribution (fnc_coverage_distrib_based)
     mean_error_3D: 13.0099
     mean_error_2D: 11.6827
    floor_det_prob: 0.6907




The software was tested with Matlab R2014b and R2016b.

Request for citation:

[1] E.S. Lohan, J. Torres-Sospedra, P. Richter, H. Leppkoski, J. Huerta, A. Cramariuc, Crowdsourced WiFi-fingerprinting database and benchmark software for indoor positioning, Zenodo repository, DOI 10.5281/zenodo.889798

