Dileep Raghav

Tradeoff Between Time and Frequency Resolution for Modulated Harmonic Signals

Date: Wednesday 1/28/04

Time-Frequency distributions (TFDs) are an important tool for the analysis of signal characteristics, providing a description of spectrum variation with respect to time. The known methods for time-frequency characterization include the frequently used and easy to implement Short-time Fourier transform (or Spectrogram), and the Cohen class of joint distribution functions, which can be customized for improved resolution, at the cost of higher �implementation complexity.

In this thesis, we compare several of the known time-frequency distributions, including the Spectrogram, Wigner-Ville and Choi-Williams distributions, in terms of their time- and frequency-resolution for modulated harmonic signals.� We also develop an improved version of the Choi-Williams distribution which offers a significantly better resolution tradeoff than either one of the standard TFDs.

Committee Members:
D. Brooks
H. Lev-Ari (advisor)
A. Stankovic