Comments on, "Improved multilevel quantization for detection of narrowband signals" by C.W. Helstrom

It is shown that optimum quantization levels may be obtained by using C.W. Helstrom's approach of the above named work (see ibid., vol.AES-24, p.141-147, Mar. 1988) but maximizing exact detection probabilities instead of approximations to it if the moment generating function of the test statistic can be expressed in rational form. Adaptive levels may be obtained for the quantizers by cell averaging, leading to constant false alarm rate (CFAR) detectors.<
>