Multiple Near-field User 3D Localization for Extremely Large Scale UPA

Application of the extremely large-scale antenna arrays and higher frequency bands in the sixth-generation communications obviously enlarge the scope of Fresnel area, which benefits the localization of users. However, the cross term between the two position parameters becomes a challenge for acquiring the three-dimensional (3D) position of a user in the Fresnel area of an extremely large-scale uniform planar array (UPA). To address this question, a generalized estimation of signal parameters via rotational invariant techniques (ESPRIT) based low-complexity and super-resolution 3D user localization algorithm is proposed, based on which a near-field channel estimation method is developed. First, two sub-UPA geometries are designed to separate the coupled term, which convert two-dimensional searching to two one-dimensional searching and greatly reduce the computational complexity. Then, the elevation and azimuth angles are separately estimated using twice the one-dimensional generalized ESPRIT algorithm, based on which the super-resolution algorithm is obtained. Especially, the elevation angles of multi-users can be estimated simultaneously, while the azimuth angles and distance are estimated sequentially to avoid the angle parameters pairing. Subsequently, the near-field channel is estimated based on the location parameters. Finally, the simulations are presented to verify the performance of the proposed algorithm.