Improvement in non-Gaussian jammers rejection with a non-linear spatial filter

A novel nonlinear generalized sidelobe canceller (GSC) is proposed and approximated with Volterra filters of increasing degree. Steady-state performance is investigated in a realistic context. It is shown analytically that a linear-cubic GSC improves non-Gaussian jammer rejection when classical linear beamformers fail (strong jammer located very near the source). This nonlinear filter is adapted to the jamming environment but does not depend on the source signal distribution; it can be estimated in the absence of the source signal and frozen when it is present, whatever the source signal distribution. This method is particularly attractive for arrays of few sensors, because, in these cases, the computational complexity remains low.<>