Dispersive covariance codes for MIMO precoding

We present a new precoder designed to utilize channel covariance knowledge at the transmitter to simultaneously achieve high capacity and an optimized error performance. These dispersive covariance codes (DCCs) utilize the linear construction of linear dispersion codes (LDCs) to spread information-bearing symbols in the time and spatial domains. It is shown that the linear framework permits the design of capacity optimal codes with desirable performance properties. Since DCCs utilize only statistical channel information instead of instantaneous estimates, feedback overhead is low, and the simple linear encoder simplifies codebook characterization. The DCCs are constructed to satisfy a capacity optimality criterion in conjunction with a pairwise error probability (PEP) constraint to improve error performance. Through numerical simulation, a two-stage DCC design procedure is shown to utilize statistical channel knowledge to provide significantly improved capacity and bit error rate performance over LDC codes in the presence of channel correlations.