| The increasing demand for high data rates and high quality communications and the lack of wireless spectrum have prompted the advent in recent years of multi-antenna wireless communication systems. However, there are two major challenges for multiple-input-multiple-output (MIMO) communications. One is to recover the transmit signals, and the other is to achieve channel capacity.; For MIMO channel equalization, in order to overcome the power enhancement problem in zero-forcing Bezout precoders/equalizers designed by using the matrix Bezout identity, we propose a FIR precoder/equalizer minimizing the residual ISI and ICI of the equalized channel under transmit power constraints/output noise power constraints. Moreover, instead of reducing ICI and ISI simultaneously, we propose a power constrained channel decoupler that applies all the design freedom to reduce ICI while ISI removal for each decoupled subchannel is left to a scalar turbo equalization.; However, the more realistic situation of imperfect channel knowledge due to channel estimation errors needs to be considered in practice. As an extension of power constrained FIR precoder design with perfect CSI at the transmitter, we consider the design of a power constrained FIR precoder for the least-favorable channel within a neighborhood of the estimated channel. The neighborhood is formed by placing a bound on the Kullback-Leibler (KL) divergence between the actual and estimated transmission systems formed as a concatenation of the precoder and channel. Instead of considering the concatenated system uncertainty, a more general neighborhood that only focuses on channel estimation errors is also used to define the channel uncertainty tolerance. It takes the form of a ball centered about the channel estimate, with a radius equal to the channel estimation uncertainty. As a dual problem of robust precoder design, a robust MSE equalizer is designed for the worst-case channel model within a neighborhood of the estimated channel model formed by the KL divergence metric. The purpose of precoders is not only to equalize channels, but also to achieve the channel capacity. For this purpose a robust power allocation scheme is proposed to achieve the channel capacity for the worst-case channel model within a neighborhood of the estimated channel model. |
