Wideband Beamspace Channel Estimation Based on Improved VAMP Algorithm

Massive multiple-input multiple-output (Massive MIMO) equipped with lens antenna arrays (LAAs), known as beamspace Massive MIMO, can focus signals into specific beams to enhance the performance and efficiency of millimeter-wave (mmWave) wireless communication systems. However, the channel estimation is extremely challenging, since the number of radio frequency (RF) chains is considerably smaller than the number of antennas. Furthermore, due to the beam squint effect in wideband mmWave systems, the channel estimation is further complicated. To tackle these challenges, we propose an improved vector approximate message passing (VAMP) scheme for the wideband beamspace channel estimation. Specifically, the channel estimation is firstly formulated as a sparse signal recovery problem, where the sparse signals can be recovered from a limited number of noisy measurements. Then, a novel shrinkage function is derived for the VAMP algorithm to enhance estimation accuracy. Finally, by replacing the original shrinkage function of VAMP with the proposed function, the improved VAMP algorithm can estimate the wideband beamspace channel with greater accuracy. Simulation results confirm the effectiveness of the proposed scheme and demonstrate its superior performance in channel estimation.