ASA 124th Meeting New Orleans 1992 October

1pUW5. Time delay estimation of noisy signals using spatial amplitude distribution analysis.

Andre M. Rog

Naval Biodynamics Lab., P.O. Box 29407, New Orleans, LA 70189

Marc W. Losh

Terry E. Riemer

Russell E. Trahan, Jr.

Univ. of New Orleans, New Orleans, LA

The estimation of time delay in signals in the presence of noise has been an historically stubborn problem. A discrete method, known as spatial amplitude distribution (SAD) analysis, is shown here to be an effective procedure for dealing with time delay estimation issues. This technique employs a window that is shifted along the data point set containing the signal (if any) and noise. The amplitudes of this windowed signal segment and the original reference signal are paired point for point, forming a two-dimensional distribution. As the window is shifted along the noisy signal a set of distributions is produced that may be analyzed independently of time. SAD map simulations involving both sinusoidal and Gaussian noise signals contaminated with Gaussian noise of varying SNR produced results indicating substantially improved performance over conventional cross-correlation techniques.