Abstract
This paper shows the practical procedure how to construct the nonlinear filter with nongaussian nonstationary random input by using the orthogonal function weighted by the probability distribution.
As for the optimizing criterion, the mean squared value of the difference between the actual and the desired output signals of the filter is taken for the convenience of mathematics. And the mean squared value of the error is evaluated quantitatively.
The technique described here is a means for circumventing difficulties in the mathematical treatment, and in the actual design of the optimum nonlinear filter, it becomes considerably simplified to decide which design is physically realizable.
