Localized Statistical Channel Modeling
Our goal is to construct a localized channel model, whose statistical distribution of angle power spectrum (ASP) is consistent with that of measured data from real environment.
Importance/impact, challenges/pain points
The number of 5G antennas has increased significantly, which improves the resolution of multi-path channels. Thus, channel model needs to be extended from one-dimensional path loss to the high-dimensional ASP model.
By building the relationship of channel vector and RSRP, sparse techniques (e.g., OMP) could be used to extract the statistics of channel gain. Data-driven and model-driven methods are used.
Key contribution/commercial implication
The localized statistical channel model could generate channel data for Simulated Reality of Communication Networks (SRCON) to predict spectrum efficiency. Then, the angles of antenna and power are tuned for network optimization.
(1) Further reduce the mean absolute error of channel model for predicting ASP of grids.
(2) Design the multi-grid scheme for channel modeling to deal with sidelobe problem.
Xi Zheng, Pu Zhang (both are from Huawei)
Qingjiang Shi, Tsung-Hui Chang, Shutao Zhang, Xinzhi Ning