Fine-grained ultrasound range finding for mobile devices: Sensing way beyond the 24 kHz limit of built-in microphones

Range finding is a core service in many applications for smart homes and cities. Recently studies have been exploring ultrasound-based range finding with commercial-off-the-shelf (COTS) mobile devices. Yet their accuracy is largely bounded by the limited sample rate (typically up to 48 kHz; i.e. only sensitive to sound below 24 kHz) of their built-in microphones and chips. In this paper, we introduce a novel coprime-sampling-based method that can enable current COTS mobile devices to achieve accurate range finding by sensing very-high-frequency (> 24 kHz) ultrasound. We show that, through smart algorithm design, the hardware limit will not be a barrier for COTS devices to derive such useful information as phase shift and power spectral density (PSD) from very-high-frequency ultrasound signals. We discuss how our method can achieve fine-grained range finding, as well as improve other applications such as motion tracking and data transport. Through evaluations of our prototype on the latest Android tablet model, we show that with an ultrasound signal at 42 kHz, our method can achieve fine-grained range finding with a median error of 1.2 mm and a median accuracy of 97.3%.