Tensor-based High-Accuracy Position Estimation for 5G mmWave Massive MIMO Systems

Publication Type:
Conference Proceeding
ICC 2020 - 2020 IEEE International Conference on Communications (ICC), 2020, 2020-June, pp. 1-6
Issue Date:
Filename Description Size
Binder2.pdfAccepted Manuscript479.18 kB
Adobe PDF
Full metadata record
Highly accurate localization is important for wire-less communications. In this paper, we propose a new tensor-based positioning method for 5G wideband mmWave massive MIMO systems. We first develop an extended multidimensional interpolation (E-MI)-based method as the preprocessing step to suppress the frequency-dependence of the array steering vectors. By using this method, the data across the whole frequency band can be processed jointly, and the high temporal resolution offered by wideband mmWave signals can be exploited. Then, we propose a parameter decoupling (PD)-based tensor multiparameter estimation algorithm. This algorithm can suppress the noises in all of temporal, spatial and frequency domains, and thus all the parameters can be precisely estimated. A simplified perturbation term (S-PT)-based method is also presented to match the estimated parameters at low complexity. Based on the quasi-optical property of mmWave signals, we propose a novel method to compute the 3D coordinates of the target. Simulation results demonstrate the effectiveness of the proposed positioning method in the end.
Please use this identifier to cite or link to this item: