| Beamforming, the task of identifying and analyzing a desired signal received amidst various interfering signals and noise, has proven to be integral to various fields, including radar, sonar, acoustics, and medical imaging. To localize the desired signal, beamforming uses a weight vector associated with a sensor array to increase the Signal-to-Interference-plus-Noise Ratio (SINR) over time. However, beamforming is often sensitive to singular covariance matrices due to finite support, as well as uncertainty in parameters, such as the signal steering vector. This creates a need for more robust methods of beamforming. Robust Adaptive Bearnforming classifies a group of adaptive beamforming methods that attempt to mitigate such uncertainties and singularities. The Loading factor Sample Matrix Inversion (LSMI) technique is one such method, though no method has been developed for determining the optimal loading factor until recently. In 2009, Li et al. proposed the use of a particle filter to optimize the loading factor and thereby improve the robustness. However, this method is computationally complex. Here, we propose the use of a marginalized particle filter, specifically the Rao-Blackwellized Particle Filter, to optimize the loading factor for the LSMI technique. An algorithm is presented and its performance is compared to that of the particle filter proposed by Li et al. |
