Robust adaptive CRLS-GSC algorithm for DOA mismatch in microphone array

In this paper, two new noise reduction algorithms are presented which are robust to the mismatch problem commonly exist between the ideal array look direction and actual speaker's Direction-Of-Arrival (DOA). Deriving a set of leakage constraints and incorporating them into the state-of-the-art Generalized Sidelobe Canceller (GSC) algorithm leads to two new noise reduction algorithms, namely Constrained Least Mean-Squares Generalized Side-lobe Canceller (CLMS-GSC) and Constrained Recursive Least-Squares Generalized Sidelobe Canceller (CRLS-GSC). Conducting several simulations, it is demonstrated that the proposed algorithms will flatten the array mainlobe and as a result, can significantly improve noise cancellation of the beamformer in the presence of DOA mismatch. Simulation results indicate that the proposed algorithms successfully attenuate the closely-placed interferers as well as they are robust to small changes in desired source DOA.