The application of two-dimensional wavelets and filterbanks for convergence enhancement in adaptive arrays for space communications

A linearly constrained wideband adaptive sensor array is described whose adaptive processor is realized with a two-dimensional tree-structured filterbank. The goal is to obtain a fast converging adaptive processor subject to a constraint on the number of adaptive weights and a requirement that the weight updating algorithm exhibit a low computational complexity. It is demonstrated via simulation that the filter bank processor utilizing the computationally simple normalized LMS algorithm, coupled with low-order multiplier-free QMF banks to provide a spatio-temporal subband decomposition, is capable of faster convergence speeds than the standard tapped-delay-line filter structure with an equivalent number of adaptive weights