INVERSE SCATTERING OF A DISCRETE LOW LOSS MULTILAYER STRUCTURE USING DIGITAL SIGNAL PROCESSING (DSP) ,TECHNIQUES 1. ' •:.!.;.•1 ;i : 'l.- • . _, •l•. . ',, .

Abstract

Recently, numerous Digital Signal Processing (DSP) techniques have been employed in solving complex electromagnetic problems. The solution of the electromagnetic inverse scattering problem for a discrete, low loss, non-dispersive multilayer system, is an ideal problem for solution with DSP techniques due to its wide range of applications, its discrete nature which adapts it easily for digital processing techniques and its highly non-linear structure. DSP solutions have the advantages of being accurate, fast (possibly real time) and computationally efficient solutions. In addition, the discrete low Joss, non-dispersive multilayer formulation allows the effects of certain phenomena (such as losses and dispersion) to appear better than any other formulation. Besides, many real life practical situations closely resemble a discrete, low loss, non-dispersive multilayer structure. Examples of such situations are radio-glaciology and Ground Penetration Radar.

In this thesis, two different DSP based inverse scattering algorithms are reviewed. The first algorithm employs the z-transform alongside with Dynamic Predictive Deconvolution (DPD) techniques. In this algorithm, the multilayer structure is modeled as an Infinite Impulse Response (IIR) filter with the properties and the losses of each layer embedded in the filter coefficients. This algorithm is based on the ability of detecting the phase of the received reflection response. The third algorithm also implements the z­ transform but is different in the sense that the low loss layered dielectric is modeled as an asymmetric discrete lattice filter which is then assembled into an asymmetric Toeplitz form solved using the asymmetric Levinson algorithm. This algorithm completely neglects the phase in the reflection response. The two algorithms are implemented and their performance compared.