![]() | Only 14 pages are availabe for public view |
Abstract In this thesis we deal with the problem of Spatial Spectral Estimation. The. aim of these techniques is that of estimating the number of signals incident on an antenna array as well as their directions with respect to the array aperture.Direction Finding systems and Radar target detection are typical applications of thesetechniques. We begin our study with conventional Fourier Transforms based algorithms such as the periodogram . These algorithms have the fundamental drawback of poor resolution which depends upon the size of the aperture as well as high sidelobe levels.Therefore, Superresolution array processing were developed such as the MUltiple Signal Classification (MUSIC) , Minimum Norm (MINORM), that exploit the eigen structure of the correlation matrix and the Minimum Variance Distortionless Response (MVDR) algorithm also known as the Maximum Likelihood (ML) and the Adapted Angular Response (AAR) algotithm. A comparative study is included to discuss the effect of various array parameters on the performance of these algorithms, namely the effects of number of sensors , SNR and angular separation.A derivation oftheresolution of the mostpopular algorithm, MUSIC, to compare it with. the resolution obtained by conventional methods.Also the effect of signal correlation is dealt with where we study the degradation in the performance of modern superresolution algorithms in correlated signal environments.An expression of the variance of the MUSIC spectrum is derived and it is compared to the Cramer Rao Lower Bound. To circumvent difficulties encountered by these methods correlated signal environments the Spatial Smoothing SPSM preprocessing scheme is introduced in chapter five . The results obtained by numerical studies showed thesignificant improvement of the performance and resolution of the spectra in the second part of this chapter, the problem of multipath propagation due to ground reflections is |