Human tracking system for indoor environments using bluetooth low energy triangulation and millimeter-wave radar sensor fusion

Abstract

This work proposes a Human Tracking System for Indoor Environments using decentralized sensor fusion that combines Bluetooth Low Energy (BLE) with Angle of Arrival (AoA) estimation and Millimeter Wave (mmWave) Radar. The experimental setup consists of a Texas Instruments IWR6843ISK 60 GHz FMCW radar producing real-time 3D point clouds, Silicon Labs BG22 BLE anchor boards with dual-polarized antenna arrays for AoA triangulation, and a mobile tag broadcasting beacons with a Constant Tone Extension (CTE). BLE-based triangulation provides high coverage tracking but suffers from large variance due to indoor multipath fading, whereas mmWave centroid estimates are robust with clear line-of-sight but degrade with range and occlusions. The proposed methodology performs coordinate alignment and time synchronization across measurements, applies a Sliding Window Median Filter (SWMF) to reject BLE outliers, and fuses the resulting tracks using an inverse-covariance Track-to-Track Fusion (T2TF) formulation. Distance dependent covariance lookup tables are employed to adapt sensor trust to spatial zones. Experimental results in a laboratory environment with annotated reference points show that fusion reduces Root Mean Square Error (RMSE) by 42.83% compared to BLE alone and by 25.40% compared to mmWave alone, yielding improved robustness across challenging indoor conditions.