|M.Sc Thesis||Department of Electrical Engineering|
|Supervisors:||Prof. Cohen Israel|
|Dr. Miriam Doron|
|Full Thesis text|
Adaptive beamforming techniques are widely used in many real-world applications such as wireless communications, radar, sonar, acoustics, and others. These techniques are effective in suppressing interference signals and noise while recovering the desired signal. In Some of the applications, wideband adaptive beamforming is required due to the wideband nature of the employed signals. One of the main approaches for implementing wideband adaptive beamforming is the coherent approach. Methods based on this approach involve a linear pre-processor which focuses the signal subspaces at different frequencies to a single frequency, followed by a narrowband adaptive beamformer such as the Minimum Variance Distortionless Response (MVDR) algorithm. The main benefits of the coherent methods over that of non-coherent methods are low computational complexity, the ability to combat the signal cancellation problem and improved convergence capabilities. In the literature, there are several methods to design focusing matrices for the coherent processing. The methods differ from each other in various features, such as the focused directions, optimality criteria, etc. In this work, we present and study a Bayesian Focusing Transformation (BFT) for coherent wideband array processing, which is robust to uncertainties at the Directions of Arrival (DOAs). The Bayesian focusing approach takes into account the probability density functions of the DOAs and minimizes the mean-square error of the transformation, thus, achieving improved focusing accuracy of the actual data over the entire bandwidth. We also treat the important issue of robust focused MVDR beamforming in order to reduce the sensitivity of the focused MVDR beamformer to errors caused by focusing errors. We generalize the diagonal loading solution and develop a robust MVDR beamformer for the coherent wideband case referred to as the Q-loaded focused MVDR wideband beamformer. Numerical results and simulations demonstrate the superior AG of the focused Q-loaded beamformer combined with BFT focusing over that of the other focusing methods. Finally, we propose and study two robust methods for coherent focused wideband MVDR beamforming. The focusing procedure introduces a frequency dependent focusing error which causes performance degradation, especially at high Signal to Noise Ratio values. The proposed robust methods aim at reducing the sensitivity of the coherent MVDR to focusing errors. The first method is based on modifying the beamformer optimization problem and generalizing it to bring into account the focusing transformations utilizing the general-rank MVDR algorithm and the second is based on modifying the focusing scheme itself.