VisBLE: Vision-Enhanced BLE Device Tracking

Wenchao Jiang, Feng Li, Luoyu Mei, Ruofeng Liu, Shuai Wang

Abstract: loT devices have evolved from providing remote connection to being an essential component of the Metaverse. The integration of loT and vision technologies has been incubating emerging applications such as vision-enhanced device tracking and remote education/medicine/maintenance. Despite the exciting vision, practical challenges include coordinate transformation, angle estimation, target mapping, and personal error. Instead of proposing yet-another localization approach, we propose a novel vision-enhanced device tracking system, called VisBLE. VisBLE takes advantage of the new localization capability introduced in BLE 5.1 and advances in vision technologies for high accuracy, robust, and intuitive BLE device tracking. There are two novel technical mechanisms: i) a rotation-based wireless localization mechanism that accurately and robustly locates the BLE transmitter in the camera coordinate and ii) a homography-based matching mechanism that identifies target BLE devices with high accuracy on the camera screen. We prototype VisBLE and deploy it on the smartphone (i.e., Nexus 5X) and development board (i.e., CC26X2 + BOOSTXL-AOA). Our results show that VisBLE outperforms the state of the art in both angular accuracy and position accuracy.