Improved energy efficiency for wireless SC MIMO through data-dependent superimposed training

In this paper, we compare the energy efficiency of a single carrier multiple-input multiple-output communications in terms of the energy per data bit required to achieve desired performance level or throughput. The comparison on energy efficiency is done between traditional time-domain multiplexed training and a more recently introduced data-dependent superimposed training. We extend our earlier single-input multiple-output system to the multiple-input multiple-output case and show how the data-dependent superimposed training based system can achieve better energy efficiency in small and diversity enabled multiple-input multiple-output links. In addition, we present analytical mean squared error results for DDST based MIMO channel estimation and propose methods to improve their accuracy when using short channel estimators.