Maximum likelihood beamforming in the presence of outliers

The problem of maximum likelihood beamforming in the presence of outliers is considered. In practice, outliers occur due to malfunctioning of sensors or as a consequence of strong impulsive noise. The performance of beamformers based on maximum likelihood or minimum mean square error type criteria is seriously degraded by outliers. One solution to combat this would be to optimally detect the failed sensors and the presence of impulses. The complexity of this direct solution increases exponentially with the number of array sensors and time samples. The authors propose an alternative method that models outliers as impulsive noise and detects impulses by using the residue of l ₁ beamformer. This technique is developed, and its efficiency is discussed