A HIGH PERFORMANCE SPECTRAL ANALYSIS TECHNIQUE FOR NON-STATIONARY SIGNALS

The premise of this SBIR topic assumes that a combination of Gabor and Wigner transforms can maintain Wigner's superior time-vs.- frequency gram resolution for characterizing non-stationary signal behaviro (relative to the time-vs.-frequency gram resolution achieved by the classical shorti-time Fourier transform) while eliminating the cross-term artifacts that are normally detrimental to grams produced by the Wigner transofrm. Under this premise, the only issue remaining to make this combination a viable time-frequency analysis technique for nonstationary signals is the development of algorithmic solutions that determine the Gabor coefficients. The co-investigators for this proposal have discovered, through a previous study of the Gabor-Wigner combination, that a modification of a high-resolution parametrric spectral analysis technqiue, origianlly developed for stationary signals, can generate time-frequency grams for non-stationary signals with better resolution than that of the Wigner and without aqny cross-term artifiacts. We propose to complete the development of ths new high-performance spectral analysis technqieu during the Phase I project, and to demonstrate conclusively tits superior capabilities in analysing complex highly non-stationary signals.