Publication: Machine Learning Based Indoor Localization Using a Representative k-Nearest-Neighbor Classifier on a Low-Cost IoT-Hardware 

Authors:
Dziubany, M. ,  Machhamer, R. ,  Laux, H. ,  Schmeink, A. ,  Gollmer, K.-U. ,  Burger, G. ,  Dartmann, G.
Book Title:
Proceedings of European Signal Processing Conference (EUSIPCO)
Pages:
p.p. 2050-2054
Date:
Sep. 2018
DOI:
10.23919/EUSIPCO.2018.8553155
hsb:
RWTH-2018-231307
Language:
English

Abstract

In order to make Internet of Things (IoT) applications easily available and cheap, simple sensors and devices have to be offered. To make this possible, our vision is to use simple hardware for measurements and to put more effort in the signal processing and data analysis to the cloud. In this paper, we present a machine learning algorithm and a simple technical implementation on a hardware platform for the localization of a low accuracy microphone via room impulse response. We give a proof-of-concept via a field test by localization of multiple positions of the IoT device. The field test shows that the recorded signals from the same source are unique at any position in a room due to unique reflections. In contrast to other methods, there is no need for high accuracy microphone arrays, however, at the expanse of multiple measurements and training samples. Our representative k-nearest-neighbor algorithm (RKNN) classifies a recording using a k-nearest-neighbor method (KNN) after choosing representatives for the KNN classifier, which reduces computing time and memory of the KNN classifier.

Download

BibTeX

Copyright © by ICE
dziubany18.pdf
This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.