Spatial modulation over partially coherent multiple-input/multiple-output channels

Communication over multiple-input/multiple-output (MIMO) channels of arbitrary coherence is considered in light of a mean square estimation error (MSEE) criterion. Earlier work in the field has focused on fully coherent channels and determined that use of a singular value decomposition (SVD) of the channel transfer function matrix can realize the capacity of the MIMO channel. More recently, research has shown that the use of arbitrary orthonormal channel excitation vectors can maximize expected capacity over fully incoherent Rayleigh fading MIMO channels. Partially coherent channels have generally been examined only in terms of their degrading influence on capacity. In this discussion, channel excitation techniques are proposed that minimize an MSEE criterion over an ensemble of MIMO channels of arbitrary coherence. The algorithms rely on only the second-and fourth-order moments of the channel transfer function. Two experiments were conducted to examine the new strategies. Using measured MIMO channel transfer function ensembles-one from an underwater acoustic channel and others from RF wireless channels-the performance of the strategies are compared. The new techniques outperform orthonormal signaling based on SINR or capacity metrics while requiring substantially less channel feedback than needed by a channel decomposition approach.