High Precision Positioning Algorithms Based on Improved Sparse Bayesian Learning in MmWave MIMO Systems

Sparse Bayesian learning (SBL) is a millimeter-wave (mmWave) positioning method that leverages the sparsity of channels to estimate parameters such as angle of arrival (AOA) and time delay for positioning. Compared to other parameter estimation algorithms, such as the Multi-signal classification (MUSIC) algorithm, Expectation–Maximization (EM) algorithm, and Space-alternating Generalized Expectation–Maximization (SAGE) algorithm, SBL demonstrates superior performance and robustness in millimeter wave scenarios. However, most existing SBL solutions only account for angle sparsity. In this chapter, we address the joint sparsity of both the angle domain and time delay domain, and propose a new two-dimensional adaptive grid refinement method to enhance the existing SBL framework. To address the grid mismatch problem common in all sparse estimation algorithms, we have also introduced a low-complexity grid evolution algorithm. Additionally, we derive the Cramer-Rao bound (CRB) for AOA, time delay, and position estimation based on the mmWave multipath signals from base stations (BS), and subsequently analyze estimation errors. Simulation results indicate that the proposed algorithm outperforms existing algorithms and approaches the CRB. Simulations using real-world datasets also confirm these findings.