Optimum MIMO-OFDM receivers with imperfect channel state information

Channel estimation inaccuracy is known to affect significantly the error performance of coded communication systems. This applies in particular to broadband MIMO channels, often considered in conjunction with OFDM such as in the IEEE 802.11n and 802.16 standards. The focus of this work is on an 802.11n compliant MIMO-OFDM communication system. Different channel estimation techniques (based on pilot symbol insertion) are considered and their relevant error performance is analyzed. More specifically, genie-aided, mismatched, and optimum channel estimation techniques are studied with special emphasis on the last one as far as concerns the relative error performance versus complexity trade-off in suboptimum implementation. It is shown that the optimum receiver can be implemented by limiting the channel processing to the dominant eigenmodes, in order to reduce the ensuing complexity. The approach followed in this work may be seen as an extension of previous results relevant to the narrowband MIMO channel.