in a clinical room used for research at the University of Utah School of Medicine's Department of Anesthesiology. Patwari reclined on a hospital bed and listened to a metronome to time his breathing so he inhaled and exhaled 15 times per minute – about the average breathing rate for a resting adult.
His breathing was measured two ways: by the experimental wireless network, and by a carbon dioxide monitor connected to his nostrils by tubes. It calculated breathing rate by measuring the amount of carbon dioxide exhaled with each breath. Patwari also tested the wireless network with no one in the hospital bed.
The study found the wireless network could measure breathing within 0.4 to 0.2 breaths per minute, an insignificant error rate given that most breathing monitors round to the nearest breath per minute, he says. If a bedridden person or baby moves, the wireless system detects the movement but cannot measure their breathing at the same time.
To decide if someone is breathing or not, the wireless system uses a computer algorithm – basically, a set of formulas. Patwari says his algorithm squares the amplitude or loudness of the signal on each link between nodes, then averages it over all 380 links. A number larger than 1.5 indicates breathing has been detected.
Patwari also measured how many nodes were required to measure breathing accurately. The minimum was 13 nodes or transceivers, while the rate of incorrect breathing measurements fell to zero when 19 nodes were used. The study also showed the height of the nodes around the hospital bed didn't significantly affect breathing measurements.
Patwari plans more research on whether different or multiple radio frequencies might detect breathing better than the one 2.4 gigahertz frequency used in the study.
He also wants to test whether the system can detect two people breathing at the same rate but not in sync – something that might make it possible to design a system that could detect not only the location of hostages in a building, but the number held together.