Channel estimation with amplitude constraint: Superimposed training or conventional training ?

Abstract

This paper utilizes a general superimposed training based transmission scheme that includes superimposed training and pilot symbol assisted modulation (PSAM) as special cases. The channel estimator of the scheme is the linear minimum mean square error (LMMSE) estimator. By taking into account errors of this method, we derive the closed-form lower bound of the data throughput under the constraint of limited amplitude for each symbol. Our study shows that with the constraint of total amplitude for each symbol, the conventional PSAM performs better in the high signal-to-noise ratio (SNR) region while at low SNR, the superimposed scheme performs better.