EXTENSIONS OF THE MINIMUM-VARIANCE METHOD

Minimum variance mean square methods with linear systems are widely us ed in many areas of signal processing such as spectral estimation, bea mforming or detection. However there is no reason to restrict these me thods to the case of linear systems. The problem is then stated and so lved in the most general context. The results are applied to some nonl inear systems and it is especially shown that for complex inputs widel y linear systems always have better performance than classical linear systems. Some extensions are also presented for detection with quadrat ic systems.