A possibility to modify the quality criterion in binary signal detection is studied. Instead of the detection error average probability, we used its root-mean-square (RMS) one containing both the error average probability and its variance. In this way the minimization of the RMS error probability makes it also possible to restrict the error probability dispersion near its mean value. As a result, the area of error probabilities is closer to zero, which indicates an improved quality of the detection. The equation of the optimal threshold value for the detection algorithm is determined. Examples with the modified quality criterion are considered and they are compared with the known quality criterion.
|
