Adaptive multichannel filtering

A linearly constrained adaptive multichannel filtering algorithm is described which is designed to adjust the filter coefficients, sample by sample, within the prescribed constraints for adaptive noise cancelling purposes. This technique is applicable to many array processing problems where in the signal is corrupted by coherent as well as random noise. By minimizing the ℓ
1
norm of the filter output with time-varying convergence functions containing a version of automatic gain control and a stochastic approximation-like sequence of constants, the algorithm is shown capable of suppressing coherent noises while leaving the signal relatively unchanged.