Near-Field MIMO Channel Reconstruction Via Limited Geometry Feedback

Near-field MIMO channels comprising line-of-sight (LoS) dominant paths can be fully described by the angle of arrivals (AoAs) and the rotation angles of both the user equipment (UE) and base station (BS), exploiting their respective antenna geometries. The resulting geometric characterization of the high-rank channels remains independent of the antenna array placement and size at both BS and UEs. Conventional near-field channel estimation requires UE-specific uplink pilots, leading to a significant pilot overhead as the number of UEs increases. In this paper, an alternative near-field MIMO channel reconstruction approach is proposed, relying on a minimal number of downlink pilots and limited geometry (angles, antenna placement) information exchange. Based on the downlink pilots transmitted from the outermost BS antennas and the prior information about the inter-antenna distance, the UEs may estimate the AoAs and rotation angles which in turn are quantized and fed back to the BS. Simulation results illustrate that just two downlink pilots suffice to characterize the channel geometry at the UEs accurately. Furthermore, an 8-bit quantization is adequate for relaying two AoAs and a rotation angle value from the UEs to the BS for full near-field LoS MIMO channel reconstruction given the known antenna placement.