LAUSR

In a range of IoT applications, edge nodes require minimal size, mass, and power/energy use. When node complexity is severely constrained, the burden of detection falls to the receiving system. In this paper we present a method of detection for RF signals motivated by wildlife radio telemetry, where miniature transmitters emit VHF pulses that are used to detect and localize animals in the field. We address the challenges of limited transmit power and large path losses that lead to low received SNR. In addition, the pulse train structure requires joint frequency acquisition, time synchronization, and detection; our approach addresses the key challenge of large uncertainty in signal parameters. Our algorithm explicitly integrates frequency acquisition and time synchronization, using nearly phase-coherent matched filtering followed by non-coherent pulse combining and a multiple hypotheses test to provide near-optimal detection performance in low SNR scenarios. We provide quantitative results showing performance as a function of SNR and signal uncertainty. When compared to the single pulse detection case as used in existing systems, our multipulse combining approach pulses improves detection performance by 4 dB, even under the conditions of temporal and frequency/phase uncertainty resulting from transmitter design and manufacturing. We also describe an efficient real-time signal processing implementation of the algorithm, including signal tracking that supports localization after initial signal detection.