Reduced-rank antenna selection for MIMO DS-CDMA channels

Receive antenna subset selection is an effective scheme to reduce the number of expensive RF chains in multi-input multi-output (MIMO) wireless systems (Gorokhov, A. et al., IEEE Trans. Inform. Theory, vol.49, p.2687-96, 2003). In order to speed up the selection process, two suboptimal antenna subset selection schemes are proposed for the downlink of direct-sequence (DS) code-division multiple access (CDMA) systems under frequency selective fading. These schemes are decremental and incremental selection algorithms (DSA and ISA, respectively), which recursively drop or select one more antenna element based on the mutual information maximization criterion for different channel realizations. Moreover, the high computational complexity of matrix inversion required by these two algorithms can be further reduced by a multistage Wiener filter (MSWF) (Goldstein, J.S., IEEE Trans. Inform. Theory, vol.44, p.2943-59, 1998). Monte Carlo simulation results show that the reduced-rank DSA not only precipitates the selection process with little capacity degradation, but outperforms the full-rank ISA by 2 dB.