Dynamic Behavior of Constrained Adaptive Array Sensor Processors.

摘要

The transient behavior of constrained wideband adaptive array sensors is examined with the goal of improving the convergence properties while simultaneously satisfying a requirement for limited computational complexity. The multichannel wideband adaptive filter is derived for the tapped-delay-line, frequency domain, lattice and Gram-Schmidt structures. The mean-square error for each structure is estimated and compared over varying degrees of difficulty with respect to the signal environment. A new multichannel wideband extension of the frequency domain structure is presented, and the performance is demonstrated to surpass that of the conventional tapped-delay-. line structure. It is shown via simulation that utilizing the Discrete Cosine Transform, the new structure is capable of converging at speeds which rival the more computationally expensive least square methods at no increase in the required number of adaptive coefficients. The multichannel wideband lattice structure is shown to perform better than the other structures and the least squares method of Sample Matrix Inversion.... Adaptive filters, Adaptive arrays, Constrained processors, Least squares algorithms, Multichannel filtering, Orthogonal signals.