Localização em ambientes internos baseada em regressão kNN utilizando beamforming

David Alan de Oliveira Ferreira, Celso Barbosa Carvalho

DOI: 10.14209/sbrt.2021.1570734205

Evento: XXXIX Simpósio Brasileiro de Telecomunicações e Processamento de Sinais (SBrT2021)

Keywords: localization kNN regression quartiles PCA

Abstract

To locate mobile devices on wireless networks, you can install three or more fixed devices, whose transmitted signals are used as location parameters. The problem with this approach is the increase in energy and monetary costs. Thus, in this article a location method using a single transmitter with multiple antennas is proposed. In this work, static tests were performed in a simulated environment of 100m2 with 361 reference points for signal power measurements. As test results, the proposed method stood out for using only 2 characteristics of signals, providing the location with an average error of 0.7979 m, standard deviation of 0.9703 m and accuracy of approximately 89% for errors below 2 m.

Download