Accession Number : ADA330136

Title :   Continuation of Signal Detection Using Polyspectra; 1 June 1991 - 31 March 1997

Descriptive Note : Final rept. 1 Jun 91-31 Mar 97

Corporate Author : TEXAS UNIV AT AUSTIN APPLIED RESEARCH LABS

Personal Author(s) : Hinich, Melvin J.

PDF Url : ADA330136

Report Date : 01 OCT 1997

Pagination or Media Count : 5

Abstract : Two methods have been developed for detecting and classifying transient signals in a noise background. One uses the third order portmanteau test I presented in "Testing for Dependence in the Input to a Linear Time Series Model," Journal of Nonparametric Statistics (1996). The test statistic is a sum of squared bicorrelations where the number of lags used is a function of the sample size. A FORTRAN program that has been written computes the standard second order portmanteau test and the third order test statistic. The second order portmanteau test statistic is the sum of squared correlations of the data in each window for the same number of lags used for the third order statistic. The program also computes the mean, standard deviation, skewness, kurtosis, 6th order cumulant, and range for the data in each window. These statistics along with the portmanteau test statistics are written to a file for graphing. Other summary statistics are computed which helps the user to identify the nature of the transient signals.

Descriptors :   *BACKGROUND NOISE, TRANSIENTS, COMPUTER PROGRAMMING, TIME SERIES ANALYSIS, FORTRAN, SIGNALS, STATISTICAL ANALYSIS.

Subject Categories : Statistics and Probability
      Acoustic Detection and Detectors

Distribution Statement : APPROVED FOR PUBLIC RELEASE