Comments on "FSK demodulation using an adaptive recursive digital filter"

In a recent paper by Werter (see ibid., vol.44, p.1822-5, July 1996), an unjustified normal frequency distribution is assumed, and then, the resulting error probability is compared with that of the traditional noncoherent frequency shift keying (FSK). However, the traditional FSK demodulators were developed because such a distribution could not be justified. This work presents a more appropriate error probability expression.