Research on High-Precision Indoor Positioning Algorithm Based on Enhanced MUSIC Spatial Smoothing Technique

Abstract

In metal-dense indoor environments (such as maglev maintenance workshops), multipath reflections generate coherent signals that cause rank deficiency in the array covariance matrix, leading to severe angle-of-arrival (AOA) estimation bias and false peaks in the conventional MUSIC algorithm. To address this issue, this paper proposes an enhanced MUSIC Spatial Smoothing Technique (MSST) for high-precision Bluetooth Low Energy (BLE) AOA positioning. The proposed method integrates three core mechanisms: (1) overlapping subarray division and covariance matrix averaging to restore signal rank characteristics; (2) dynamic noise subspace separation based on an eigenvalue distribution threshold to suppress residual coherent interference; and (3) an optimization-driven aperture configuration to balance angular resolution with computational complexity. The algorithm was experimentally evaluated using an nRF52833-driven single-anchor receiver equipped with a 12-element uniform linear patch array in a real metal-dense scenario, benchmarked against high-precision laser tracker ground truth. Experimental results show that, compared with conventional MUSIC, MSST reduces the −3 dB–3 dB mainlobe width from 22° to 0.82° (a 96.3% compression), enhances the sidelobe suppression by 12.56 dB (reaching −34.54 dB), and reduces off-diagonal covariance energy by 62%. The eigenvalue separation ratio λ1/λ2 increases from 1.8 to 5.3, and the overall positioning root-mean-square error (RMSE) improves from 1.20 m to 0.28 m. These results validate MSST as an effective and high-precision solution for indoor positioning under severe multipath conditions.