Robust FIR precoder design with imperfect channel knowledge for broadband MIMO wireless systems

This paper considers a robust FIR precoder design problem for frequency-selective multiple-input multiple-output (MIMO) wireless systems under transmit power constraints with imperfect channel knowledge at the transmitter. The robust design problem requires finding the optimal power constrained FIR precoder for the worst-case channel model within a neighborhood of the estimated channel. The neighborhood is formed by placing a bound on Kullback-Leibler (KL) divergence between the actual and estimated channels. We show that the optimal power constrained FIR precoder designed for the estimated channel under the assumption of perfect channel knowledge is robust when channel uncertainties are measured by the KL divergence "metric".