Frequency and Polarization Diversity in through-the-wall Breath Detection

TitleFrequency and Polarization Diversity in through-the-wall Breath Detection
Publication TypeConference Paper
Year of Publication2015
AuthorsNarbudowicz, A, Ammann, MJ, Ruvio, G, Dell’Aversano, A, Solimene, R
Conference NameProgress In Electromagnetics Research Symposium 2015
Conference Start Date06/07/2015
PublisherProgress In Electromagnetics Research
Conference LocationPrague, Czech Republic

In this paper a novel through-the-wall (TTW) breath detection system is studied in terms of different operating frequencies and signal polarization diversity for improved accuracy. The measurement set-up is organized as in Figure 1(a) and includes two sets of circularly-polarized patch antennas (Figure 1(b)) fed by a Vector Network Analyzer (VNA) placed one meter away from a conventional 12-cm thick wall. A subject breathing normally is positioned at a distance of one meter from the other side of the wall with the chest oriented in the direction of maximum radiation of the antenna (1.5 meter from the ground). The return loss of the antenna is acquired at the frequencies 1,575 and 2.575 GHz for vertical, horizontal and Circular-Polarization (CP) of the antenna system, respectively. This large dataset allows a deep investigation on the effects of frequency- and polarization-diversity on the accuracy of the vital sign detection. In particular, measurements taken for different polarizations allow an understanding of the effects of ground bouncing and signal rotation.
The antennas used for the study are CP patches, which offer high gain at minimum complexity. Measurement at 1.575 GHz were conducted using two antennas, located next to each other: one rightand one left-hand circularly polarized. For 2.575 GHz measurement, where the spacing between antennas is more critical due to shorter wavelengths, a dual circularly polarized antenna [1] was used. This allowed to achieve perfect colocation of both polarizations.
As for the breath detection methods, different algorithms are evaluated. In particular, a classical Fourier based method is compared with a novel version of the MUltiple Signal Classification (MUSIC) algorithm.

Full Text