Michael A. Staiano
Staiano Eng., Inc., 1923 Stanley Ave., Rockville, MD 20851
A correction scheme has been proposed which consists of repetitive measurements of the source-signal-with-background and background noise alone than the computation of a signal estimate and prediction interval. The procedure assumes that both the source of interest and background noise are: uncorrelated, normally distributed, random processes which are stationary over the duration of the measurements. For useful results, the numbers of measurement must be selected to provide for a calculated confidence interval which acceptably contains the prediction errors. These requirements are strongly influenced by the variability of the measured parameters. For relatively low background noise situations, the technique is useful primarily for quantifying expected measurement confidence bounds. Application of the procedure to a field measurement situation with mean source band sound levels ranging approximately from 2 to 15 dBA re: mean background band sound levels gave a 99.5%-confidence upper-bound margin of 0.3--1.3 dB with 17--19 measurements per band.