Key points are not available for this paper at this time.
A minimax formulation is proposed for the problem of designing robust Wiener filters for the situation of uncertain signal and noise spectra. Several results relating to solutions to this formulation in terms of least-favorable spectral pairs are presented. An analogy is drawn between a particular case of this problem and the problem of robust hypothesis testing, and a result is given concerning the relationship between solutions to these two problems.
H. Vincent Poor (1980) studied this question.