Transient detection in non-Gaussian noise by wavelet packet transform

An algorithm based on wavelet packet transform is proposed for detecting a deterministic transient signal with unknown parameters in non-Gaussian noise. Expressions for probability of detection and probability of false alarm are derived, and performance results are presented in the form of ROC plots for detection of damped sinusoids in different noise distributions. It is shown that, for Gaussian noise, the performance of the proposed WPT detector is much better than that of the generalized likelihood ratio test (GLRT)-based detector. Variation of performance of the WPT detectors with different noise distributions and detector parameters is also investigated.