|
The received signals in Sonar, Radar, and many other
surveillance systems are generally expressed in non-stationary
non-Gaussian stochastic processes, and they must be processed
by appropriate time-frequency analysis methods in order to
obtain correct information, especailly for the weak (low or
negative SNR) non-stationary transient signals. The objective
of this thesis is to search for pertinent processing algorithms
for detection and identification of transient signals using
higher-order spectra (HOS), and to develop technique which can
be applied to underwater acoustic signal processing or the
Sonar system. There are three motivations behind the use of HOS
in signal processing, namely, to extract information due to
deviations from Gaussianity, to identify non-minimum phase
signals and to detect and characterize nonlinear mechanism.
Thus HOS play a key role in growing applications of engineering
and science. Higher-order spectra are investigated in this
thesis, including the conventional indirect and direct methods
of HOS estimation via moment and cumulant, and Wigner higher-
order spectra (WHOS). Different transient signals contaminated
by various noises are detected with simulation and analysis
using MATLAB. In order that the proposed algorithms are
feasible for digital hardware implementation in the near
future, so LabVIEW is used to built up the algorithms.
|