Reduced-Complexity Decimeter-Level Bluetooth Ranging in Multipath Environments

In this paper, we investigate enhanced super-resolution range estimators with decimeter-level accuracy for multi-antenna and multipath Bluetooth systems. To enhance the traditional MUSIC range estimator for a two-way frequency-hopping Bluetooth channel model, we apply forward-backward averaging and bandwidth extrapolation using Burg’s algorithm to improve ranging accuracy, which is limited by the used Bluetooth bandwidth and the quality of the estimated sample covariance matrix. For the multi-antenna case, we compare the Summed Antenna Processing and Individual Antenna Processing methods to process multiple-antenna Bluetooth channel measurements and enhance the range estimation accuracy compared to the single-antenna case. In addition, we investigate a sparsity-aware range estimator which exploits the sparsity of Bluetooth channel impulse response and achieves comparable ranging accuracy to the enhanced MUSIC estimator but at a much lower computational complexity. We apply the greedy Orthogonal Matching Pursuit algorithm to heuristically solve the sparsity-constrained optimization problem for Bluetooth ranging. Furthermore, we evaluate the computational complexity of our investigated Bluetooth range estimators with two complexity-reduction techniques to further reduce the complexity of MUSIC range estimator. Moreover, we analyze the Cramer-Rao Lower Bound (CRLB) on unbiased range estimation using the frequency-hopping Bluetooth channel model and derive a new insightful CRLB expression for a two-path channel model. Finally, we evaluate the Root-Mean-Square Error and Empirical Cumulative Distribution Function performance of our investigated range estimators both on simulated and real-world Bluetooth data that we collected in line-of-sight (LOS) and non-line-of-sight (NLOS) multipath scenarios. Our proposed enhancements on the range estimators improved the ranging accuracy by 58% for our collected Bluetooth data.