Measurement-Based Channel Modeling for Spatial Non-Stationarity With Multipath Components

Due to the deployment of extremely large antenna array (ELAA) in the sixth generation (6G) networks, the so-called “spatial non-stationarity” effect becomes increasingly prominent. Spatial non-stationarity, characterized by the variant complex attenuation coefficients across the antenna array, has been observed in many measurements. Therefore, it is urgent to establish a statistical channel model to accurately model and reproduce this channel characteristic. In order to solve the problem of high-order path number when utilizing space-alternating generalized expectation-maximization (SAGE) algorithm under ELAA channels with spatial non-stationarity, in this work, the Optimized Searching-Strategy aided Space-Alternating Generalized Expectation-maximization (OSS-SAGE) algorithm has been proposed. By utilizing this algorithm, the multipath components (MPCs) are extracted to establish a channel model that incorporates both spatial non-stationary characteristics (SNC) and near-field effects. To maintain the compatibility with the structure of the conventional spatial channel model (SCM), the spatial non-stationarity is modeled by the constructive and destructive addition of the MPCs. Thus, the existing SCM procedures and definitions such as cluster can be employed without significant modifications. In order to present the procedure of channel modeling and evaluate the performance of the OSS-SAGE, extensive channel measurement campaigns have been conducted in a hall scenario. The measurement system configures a physical 32-element Rx linear array and a 32 × 32 Tx planar array at 10 GHz. A combination of a radio switch and a positioner configures the Rx planar array, balancing time and material costs. Meanwhile, the performance evaluation of OSS-SAGE indicates that the proposed algorithm surpasses SAGE in extracting overall channel characteristics and accurately reproducing the observed SNC. Finally, an exemplary statistical channel model has been established. This work provides a reference for future research on ELAA in communication systems, encompassing data acquisition, parameter estimation, and channel modeling.