Preservation of information in matrix filtering of a coherent signal correlated with noise
Derrill J. Bordelon · The Journal of the Acoustical Society of America · 1975
With the concept of mutual information about a coherent signal in additive noise, where signal and noise are assumed jointly Gaussian, it is shown that information is preserved, in a trivial way, by block diagonal matrix filters of full rank and by weighting and summing matrix filters of rank 1 which are identical, within a multiplicative scalar, with the following matrix estimator filters: (1) maximum signal−to−noise ratio estimator, (2) best linear unbiased estimator, (3) minimum mean−square error estimator, (4) maximum liklihood estimator, and (5) maximum a posteriori estimator. Subject Classification: 60.XX.