Multipath Extraction Based UL/DL Channel Estimation for FDD Massive MIMO-OFDM Systems
Wei Ji,Feng Zhang,Ling Qiu
DOI: https://doi.org/10.1109/access.2021.3081497
IF: 3.9
2021-01-01
IEEE Access
Abstract:In this paper, we propose a multipath extraction based uplink/downlink channel estimation scheme for wideband massive multiple-input multiple-output (MIMO) systems, where orthogonal frequency division multiplexing (OFDM) is adopted and the beam squint effect is considered. Firstly, we investigate the spatial- and frequency-wideband effects and obtain the relationship between angle-delay information and coordinates of on-grid paths. Secondly, according to the spatial- and frequency-wideband effects, we design a pilot pattern in uplink training to ensure that exact paths can be acquired from limited potential paths. Combined with pilot pattern design, a modified density-based spatial clustering of applications with noise (M-DBSCAN) approach is proposed to acquire the on-grid potential paths. Potential paths are used for initialization in uplink channel extraction. Due to the fact that coordinates of only a few potential paths are close to those of true paths, the uplink multipath extraction problem can be regarded as an off-grid sparse signal reconstruction problem. To deal with this issue, an off-grid sparsity adaptive matching pursuit (OSAMP) based compressive sensing method with low computational complexity is proposed. In OSAMP, angle-delay information as well as path gains of exact paths are accurately estimated, which are further used for accomplishing the uplink channel reconstruction. Angle-delay reciprocity between uplink and downlink channels is exploited, and downlink channel estimation is accomplished with the help of uplink multipath extraction. Downlink pilots are only used for downlink path gain estimation, therefore, the downlink training overhead will be significantly reduced. Simulation results are provided to demonstrate the effectiveness of our proposed low computational complexity channel estimation method.
computer science, information systems,telecommunications,engineering, electrical & electronic