Combining Space-Time Block Codes and Multiplexing in Correlated MIMO channels: An Antenna Assignment strategy. Hilde Skjevling, David Gesbert, Nils Christophersen Department of informatics University of Oslo hildesk@ifi.uio.no, gesbert@ifi.uio.no, nilsch@ifi.uio.no Abstract Multiplexing and space-time coding are compet- ing ways of extracting capacity out of MIMO wireless systems. We address the problem of finding an optimal combination of these ap- proaches over a MIMO array when a priori know- ledge about correlation is available at the trans- mitter. Our approach is in the form of an optimal spatial assignment of antennas, over which to multiplex space-time coded symbol blocks, in the highly practical case when correl- ation is not uniform across all antenna pairs. 1 Introduction One way to improve data rate and transmission reliability over wireless links is through the use of Multiple-Input Multiple-Output (MIMO) sys- tems (see e.g. [4] for a recent tutorial). Diversity-oriented transmission through space-time coding (STC) [7] and spatial mul- tiplexing (SM) [3] are two different, so far competing, approaches to exploiting the spatial dimension offered by MIMO systems. The former uses the antennas to combat Rayleigh fading, while the latter uses spatial degrees of freedom to increase data rate by sending independent symbol streams simultaneously. The trade-offs between those approaches are only beginning to be understood [4]. While it is clear that diversity schemes yield diminishing returns when increasing the num- ber of antennas [2], it is also known that an SM scheme with a simple (e.g. linear) receiver lag in performance because of a lack of diversity [5]. We address both of these open problems by attempting to combine the schemes. Previous work has been carried out in the case where the desired combination is to switch between STC and SM over time [6]. There, the proposed scheme exploits the fact that STC is sensitive to total channel matrix energy, while SM performance depends on the channel eigenvalue spread. Our approach will be different, as we focus on the problem of switching between SM and STC over space. To our knowledge, this problem has not been addressed before. We propose simple algorithms allowing us to generalize the work of [6], both for the case where instantaneous full channel feedback is available to the transmitter and for the case where only long term correla- tion statistics are known. We show the perform- ance gains with bit-error rate (BER) simulations using a realistic channel model. Notation x, x and X denote a scalar, a vector and a mat- rix, respectively. x * is the complex conjugate of the scalar. For vectors and matrices, x * denotes the complex conjugate of each element, while x T and x H are the transpose and the conjugate transpose. X # is the pseudo-inverse of X. √ X is the hermitian square root of X. 1