PIR sensors: characterization and novel localization technique

Abstract

Pyroelectric Infra-Red (PIR) sensors are used in many applications including security. PIRs detect the presence of humans and animals from the radiation of their body heat. This could be used to trigger events, e.g., opening doors, recording video, etc. PIRs are used widely because of their low power consumption. Hitherto, PIR sensors were used for binary event generation -- human/animal present or not-present. At the same time simple binary output hinders the use of PIR sensors in a wide variety of sophisticated applications. In the literature, we find limited characterization of analog output from PIR sensors that could provide much more information. We built a simple array of PIR sensors and packaged them in a tower. We used two sets of four PIR sensors and tapped their analog signals after amplification. Our major contribution is the characterization of analog signals from the PIR sensors. We describe many interesting aspects obtained from the analog signals, which have not been explored until now. We also show their correspondence with the range, speed and size of the moving object. Using the characterization of PIR sensors analog data as well as simple binary decisions from these PIR sensors, we: (i) classify moving object with high precision; and (ii) localize the moving object. The major incentives are low operating power compared to WSNs. We achieve 30 cm accuracy in 80% of the times, when ranging up to 5 m. Over multiple experiments for different persons in the range 1--10 m, we show that the error probability for localization is 0.08 at moderate distances (around 5--6 m). Our work will help in designing better detection and application triggers using PIR sensors in the near future. We believe that this work will open up new avenues in the development of new applications with PIR sensors.