The benefits of multiple antennas at both the transmitter and the receiver in a wireless link consist of increased reliability and data rates. As the signal-to-noise ratio (SNR) approaches infinity, there exists a tradeoff between reliability (diversity) and data rate (multiplexing). Such an asymptotic tradeoff is applicable only to high SNRs, high data rates, correspondingly low error rates, and ideal channels.
In this talk, a new finite-SNR (nonasymptotic) framework is presented to analyze the diversity-multiplexing tradeoff in multiple-input multiple-output (MIMO) systems. Finite-SNR multiplexing and diversity gains are defined in the context of rate adaptation. Outage probability bounds, derived from random matrix theory, are used to estimate the diversity gain as a function of SNR and multiplexing gain. Diversity gains at realistic SNRs are significantly lower than high-SNR asymptotes. Relations among the finite-SNR framework, spectral efficiency, and minimum mean-square error (MMSE) estimation will be explored.