Broadband GSC beamformer with spatial and temporal decorrelation

Broadband adaptive beamforming algorithms based on the least mean square (LMS) family are known to exhibit slow convergence, if the input is correlated. In this paper, we will utilised a recently proposed broadband eigen value decomposition method to provide strong spatial decorrelation, while at the same time reduces the subspace in which the beamforming algorithm operates. Additional temporal decorrelation is gained by operating the beamformer in oversampled filter banks. Hybrid structures which combine both spatial and temporal decorrelation demonstrate to provide faster convergence speed than the normalised LMS algorithm or either of the decorrelation approach on its own.