Wavelet-based signal processing of electromagnetic pulse generated waveforms

Abstract

This thesis investigated and compared alternative signal processing techniques that used wavelet-based methods instead of traditional frequency domain methods for processing measured electromagnetic pulse (EMP) waveforms. The primary focus of the research was equalization and filtering techniques for processing EMP signals in additive white noise. Signal equalization was conducted at the sub-band level through the use of Infinite Impulse Response (IIR) filters and channel response characteristics. A brief investigation of signal de-noising through wavelet thresholding was also conducted. This thesis also addressed and provided viable methods for signal extraction and DC bias removal for a given measured EMP waveform. The mean squared error is used as the basis for the comparison of the effectiveness of the equalization algorithm. It was found that wavelet techniques provided results that were as well or better than traditional Fourier techniques. In systems with additive noise , wavelet-based techniques exceeded the performance of the Fourier-based methods and surpassed them when de-noising techniques were used.