MIMO channel estimation with dimension reduction

Abstract

The number of channel estimates that have to be estimated in multiple-input multiple-output (MIMO) system is in general much larger, than in a single antenna communication scheme. This leads to lower signal to noise ratios (SNR) of the channel estimates if a constant pilot power independent from the number of transmit antennas is assumed. The use of long-term spatial channel characteristics can improve channel estimation for MIMO wireless systems. Separating the signal and the noise subspace followed by a dimension reduction can significantly reduce additive noise on channel estimates. This leads to improved channel estimation, especially for MIMO systems with high numbers of antennas, and to lower pilot power requirements.