Abstract: We address the problem of frequency-selective channel estimation and symbol detection using superimposed training. The superimposed training consists of the sum of a known sequence and a ...