Channel estimation for an SVD-MIMO System

Systems using singular value decomposition (SVD) over multiple-input multiple-output (MIMO) channel require the channel state information (CSI) at the transmitter and the receiver. The receiver retrieves CSI through estimation of pilot symbols from which it calculates an optimum decoding matrix. However, the pilots are affected by noise and provide an incorrect CSI. In a slow fading channel, tracking the channel in time can improve the estimated CSI. The tracking can occur on the CSI itself, or directly on the coefficients of the MIMO decoding matrix. It is shown that tracking of the CSI provides a better estimate of the MIMO decoding matrix than direct tracking of matrix itself.