BASAT: BLE AOA-Based Spatio-Temporal Acoustic TOA Estimator

In severely reverberant indoor environments, acoustic Time-of-Arrival (TOA) localization on commercial smartphones is prone to multipath-induced bias and unreliable ranging. This article presents a Bluetooth Angle-of-Arrival (AOA)–assisted spatiotemporal multipath suppression framework for acoustic TOA estimation. First, a soft conic geometric prior is constructed from AOA, and the associated angular uncertainty is mapped to the distance–time domain to enable temporal gating that preserves the direct-path component while suppressing late reverberation. Second, a Gaussian-mixture intensity prior is established using the Received Signal Strength Indication (RSSI). By inverting the Log-Distance Path Loss (LDPL) model with variance propagation, a distance–time distribution is derived and intersected with the AOA reachability bound to form a Bayesian Geometric TOA window (BG-TOA), followed by soft gating to attenuate outlier paths. Finally, under BG-TOA constraints, the joint covariance is updated online and eigenvector-based spatial filtering is performed to enhance the direct path and suppress multipath clusters, yielding the earliest consistent TOA estimate. Field experiments in an office and a large indoor space achieve mean static and dynamic localization errors of 0.11 m and 0.13 m, respectively, substantially outperforming single-sensor baselines and conventional fusion methods. The proposed framework is fully compatible with existing smartphone hardware, offering low cost and deployment simplicity.