Computationally efficient robust widely linear beamforming for improper non-stationary signals

In this work, we introduce a computationally efficient Kalman-filter based implementation of the robust widely linear (WL) minimum variance distortionless response (MVDR) beamformer. The beamformer is able to achieve the same performance as the recently derived robust WL MVDR beamformer, but avoids the computationally burdensome solution based on a second order cone programming (SOCP), and exploiting the recent Kalman-based regular robust MVDR beamformer, extends this to also allow for non-circular sources and interferences. Numerical simulations illustrate the achieved performance.