An Accurate Channel Prediction Method for Massive MIMO-Based LEO Communications

Abstract

As a promising component of the beyond fifth gener-ation (B5G) and forthcoming sixth generation (6G), massive multi-ple input multiple output (massive MIMO)-based low earth orbit (LEO) communication is facing unprecedented serious doppler frequency shifts and delays due to its fast relative moving speed between the transmitters and receivers, which makes the accurate channel state information (CSI) hard to obtain. In order to solve this problem, we introduce an improved channel prediction method, named Prony-based spatial-delay domain (SDD-Prony) prediction, it not only achieves an accurate CSI acquisition for fast-speed relative motion massive MIMO-based LEO commu-nications, but also greatly reduces the computational complexity. Besides, we find that the prediction error of our method converges to zero when the number of antennas and bandwidth growing large, provided that only two sufficiently accurate channel samples are needed. The validness of our theoretical analysis is verified by numerical results, and the simulations also further demonstrate the effectiveness of our method.