Training sequence optimization: comparisons and an alternative criterion

Two previously proposed training sequence optimization techniques for channel estimation are compared. One method is based on a frequency-domain (FD) based channel estimation method and the other is based on a time-domain (TD) channel estimation technique. The FD method produces a lower complexity search strategy, but does not always result in the optimal training sequences in terms of the mean-squared channel estimation error. A proof of the superiority of the TD method over the FD method is presented. Based on the proof, an alternative search criterion is proposed, which, in general, provides equivalent or better performance than the FD method while still enjoying the low search complexity.