A robust RLS-based generalized sidelobe canceller using spatial regularization

This paper develops an adaptive beamformer that is robust to uncertainty in source direction-of-arrival (DOA) and is able to track changing DOAs. In the proposed method, a network of generalized sidelobe cancellers (GSCs) are formed from a diffusion recursive least squares (RLS) criterion. The resulting adaptive beamformers that point at a set of candidate DOAs are smoothed via a spatial regularization . The RLS optimization is implemented by the hardware-efficient QR decomposition (QRD) structure. The best beamformer is then selected from the network using the minimum variance criterion. The mismatch analysis theoretically shows that the source signal reduction caused by DOA mismatch could be alleviated by the spatial regularization. Simulations are carried out to evaluate the performance of the proposed algorithm, which outperforms the conventional algorithms using a single beam in the presence of sensor uncertainties. Moreover, the tracking capability of the proposed algorithm has been tested for time-varying DOAs.