Ultra Low-Power, Lightweight, Probabilistic RSS-Based Path Reconstruction: A System for Landscape-Scale Bee Tracking

Tracking small, power-constrained subjects like flying insects at a landscape scale is challenging because traditional Global Navigation Satellite Systems (GNSS) are too heavy and power-hungry. The authors propose a novel, ultra-low-power localization system that avoids GNSS by using Received Signal Strength (RSS) measurements. Instead of requiring dense sampling to find signal peaks, the system uses rotating high-gain transmitters and a probabilistic model to infer the Angle of Arrival (AoA) from very sparse data bursts. The receiver, weighing only 38mg, logs these signals, and the full movement path is later reconstructed using a Gaussian process and doubly stochastic variational inference.

The researchers demonstrate that their method is highly efficient, achieving approximately 15m accuracy with as few as 3 RSS measurements per burst, consuming less than 180uW. By increasing the number of measurements to 10 per burst, accuracy improves to approximately 10m at a cost of less than 600uW. The authors validate the system through ground-truthed field experiments and demonstrate its utility by tracking the return flights of displaced buff-tailed bumblebees (Bombus terrestris) in a complex, hilly landscape.

This system provides a viable path for behavioral ecology studies that were previously impossible due to tag weight and battery limitations. By enabling the tracking of insects as small as bumblebees, the method allows researchers to observe natural navigation and foraging behaviors in complex environments without the energetic burden of heavier transceivers or the power demands of GNSS. The use of archival data logging makes it particularly suitable for studies where subjects can be recaptured.