Localisation of sheep using Bluetooth Low Energy (BLE) proximity

Application
Useful information regarding animal behaviour and social contacts can be gathered through monitoring animal location, proximity, and movement. However, many technologies are too expensive or difficult to apply in sheep systems, whereas BLE could present a cheaper and less power intensive option.

Introduction
This study aimed to assess the potential of utilizing BLE as a means of sheep localisation within a field environment. Small beacons (c 14 g) on sheep transmit their identity to static logging and communicating readers either at fixed locations or fitted as collars to other sheep.

Materials and methods
A calibration study using 6 BLE readers and 8 BLE beacons was conducted within a field setting to determine the relationship between a beacon's reported signal strength index (RSSI) and its distance from a BLE reader. Beacons were tested at distances of 1–128 m at log intervals, with a total possible 240 readings per interval. This resulting regression equation, log10 (Beacon Distance from reader) = −2.49 + (RSSI × −0.05) (R2Adjusted = 0.7529, F(1,1375) = 4193, P < .0001), was used to develop a distance prediction equation using RSSI. Two localisation studies are reported showing how a multilateration approach could be used to estimate beacon positions based on combining RSSI from 3 or more readers. This approach was tested firstly in a static beacon study with 6 BLE readers attached to fence posts (2 along the width, and 4 along the length) of a paddock and 16 BLE beacons attached to posts within the paddock in a grid-like array ∼60 × 90 m. This was followed by an on-sheep study where weaned lambs were fitted with both BLE beacons and separate Global Navigation Satellite System (GNSS) devices. In this instance 9 BLE readers were attached to fence posts around 2 adjoining paddocks (∼1.4 Ha); 4 along the length of each outer fence line, and 1 located at the gate between paddocks. Data from 3 stationary lambs is presented to validate the method, comparing location from the GNSS unit with multilateration estimations from the BLE data.

Results
In the static beacon study, 12 of the 16 beacons were located using the multilateration methodology. Estimated beacon locations ranged from 6.53–32.38 m from beacon GNSS locations, with a mean difference of 28.5 m. In the on-sheep study, estimated locations were generated for all 3 lambs. The distances between the estimated location and the lambs GNSS location was 18.38 m, 35.12 m, and 3.91 m for each lamb respectively. Giving a mean difference of 19.14 m between the predicted BLE generated locations and the GNSS locations.

Conclusions
Analysis showed BLE could be utilised for livestock localisation in settings with high densities of BLE readers.

Acknowledgments
This study is part of the TechCare Project which has received funding from the European Union’s Horizon 2020 research and innovation programme (grant agreement No 862050).