The USC Andrew and Erna Viterbi School of Engineering USC Signal and Image Processing Institute USC Ming Hsieh Department of Electrical Engineering University of Southern California

Technical Report USC-SIPI-243

“Analysis of Real Sonar Signals Using Non-Linear Signal Processing”

by Sam Heidari and Chrysostomos L. Nikias

September 1993

Various sequential real sonar data files are analyzed through different signal processing algorithms. Data files are divided into two groups: files consisting of a signal of interest (SOI) with additive noise and files consisting of pure noise. The purpose of these analyses is to determine if a given data record contains the SOI. Methods of Power Spectrum Density estimation, Higher-order Spectrum estimation and Local Intrinsic Dimension (LID) estimation are applied. A new method is introduced in LID estimation using higher-order off diagonal cumulants and shown to be more consistent in categorizing the SOIs. Further work needs to be done using Higher-order spectra-based time-frequency techniques for non-stationary analysis due to the time-varying nature of the signals.

To download the report in PDF format click here: USC-SIPI-243.pdf (1.7Mb)