Mean Estimation of Random Process with Aid of on-off Detectors

This paper presents a maximum likelihood method (ML) of mean estimation of a random process when the measurement of the exact value of process is not required. The estimation is accomplished with an array of on-off detectors. It is shown, that there exists an optimal value of standard deviation which gives the maximum value of accuracy for the worst case, i.e., when the measured mean value is exactly between detectors. Comparison of the proposed method (PM) with a commonly used method (CM) indicates that for a wide range of standard deviation of the considered process the accuracy of both methods is almost equal.