Methods and apparatus for an adaptive blocking matrix

摘要

A noise correlation statistic is estimated between two input signals 202,204. If the statistic meets a predefined condition then it is used to estimate an inter-sensor model h[n] that relates desired signal components present in the input signals. Otherwise, the model is estimated using a constrained statistic that is derived from the noise correlation statistic. The estimated inter-signal model 402 may then be used in an adaptive blocking matrix 210 to estimate noise signal b[n]. This noise signal may then be used (130, fig. 1) to perform adaptive cancellation on a beamformed signal derived from the input signals. More than two input signals may be treated similarly by estimating pairwise inter-sensor signal models. The condition on the noise correlation statistic may involve a maximum energy threshold of the normalised noise cross section, and the constrained statistic may be obtained by rescaling the noise correlation statistic by its norm.