Forgetting Factor Analysis of Least Squares Channel Estimation with Massive Number of Antennas for High-speed Railways

The performance of the exponentially weighted least squares multiple input multiple output (MIMO) channel estimation for high-speed railways (HSR) is assessed where the base station employ massive number of antennas. The channel is Rician flat fading and its time-variation is modeled as a first order autoregressive (AR1) process. The channel estimation mean-square error (MSE) is calculated at the train for different forgetting factors, Doppler rates, and number of antennas. Under the assumption that the train has no apriori location knowledge, it is shown that the forgetting factor which results in a lower MSE is highly influenced by the number of antennas, but barely affected by the Doppler rate.