Design of linear dispersion codes for practical MIMO-OFDM systems

In this work, we consider the problem of designing space-time-frequency linear dispersion (LD) codes in wideband multiple-input multiple-output (MIMO) antenna systems employing orthogonal-frequency-division-multiplexing (OFDM). Three criteria are presented and discussed in detail, which involve: 1) minimizing the average uncoded block error rate, 2) maximizing the ergodic mutual information, and 3) a two-step procedure considering the optimization of the ergodic mutual information as well as the average uncoded block error rate, respectively. For any set of subcarriers, any number of OFDM symbol intervals, any number of transmit/receive antennas and any statistical fading channel model, the corresponding optimized LD code matrices are numerically computed via a stochastic gradient descent algorithm assuming either maximum likelihood or linear zero-forcing decoding. As an application of the proposed criteria, code design examples are presented for practical next-generation communication systems which operate over a realistic 3GPP spatial channel model. The robustness of the resulting codes over several operating conditions is comprehensively demonstrated via numerical simulations.