| Most engineering systems in communication, control, and signal processing have been developed under the assumption that the interfering noise is Gaussian. In many real applications, the prevailing noise environment cannot be represented by a Gaussian distribution, with one important category of non-Gaussian noise being impulse noise.; The objective of this thesis is to propose adaptive robust filtering techniques to clean the impulse noise in a correlated, possibly nonstationary, environment. The most significant advantage of the proposed filtering is that in contrast to currently available techniques it doesn't require any iterative and/or batched processing. This is a direct result of the proposed robust-minimum-mean-square-error (RMMSE) estimation method.; The presented techniques are applicable to a wide range of problems, such as detection, power spectral estimation, and jamming or clutter suppression in impulsive environments. As a direct application of the proposed filter, the adaptive robust power spectral estimation method is proposed. Additionally, the proposed technique is applied to the problems of suppressing both narrowband interference and impulse noise in direct sequence spread-spectrum (DSSS) systems. |
