Beamforming of a distorted line array in the presence of uncertainties on the sensor positions

Array processing based on a linear assumption implies degradations in the source bearing estimation when the array is not linear, which is the usual case in real-world problems. Assuming only that the array is planar (but not necessarily quasilinear), we describe a method to include in an adaptive way information about the sensor locations into the array processing, in order to take into account its geometrical shape. This information can be the relative positions of the sensors given by a mechanical device included in the array. The relative sensor positions can also be estimated using a source with a known bearing, a cooperative target for example. Simulation results in a real underwater noise sample show that these two cases lead to important improvement with respect to the classical array processing, based on a linear assumption, provided that the signal-to-noise ratio of the known source is high enough (around 10 dB in our example).