ASA 127th Meeting M.I.T. 1994 June 6-10

4pUW16. Active detection prediction using higher-order moments.

Lisa A. Pflug

Naval Res. Lab., Stennis Space Center, MS 39529-5004

George E. Ioup

Juliette W. Ioup

Dept. of Phys., Univ. of New Orleans, New Orleans, LA 70148

For some signals, higher-order moment detection methods have been shown to perform better than conventional methods for stationary power signals and deterministic finite-energy transients. The latter case has been investigated by the authors using Monte Carlo simulations for both active and passive detection and recently, using theoretical prediction formulas for passive detection. The corresponding prediction formulas for active detection are presented in this paper. From knowledge of the first four signal moments the signal-to-noise ratio (SNR) at the minimum detectable level is obtained for the cross correlation, bicorrelation, and tricorrelation central ordinate detectors for an arbitrary probability of false alarm. The prediction formulas require that the noise have Gaussianly distributed correlation central ordinate values. Comparisons are made between the theoretical predictions and results of computer simulations for a representative set of test signals in Gaussian noise. [Research supported by Office of Naval Research, Program Element 61153N, with technical management provided by NTLSSC.]