Improved Frequency-selective Filters

This paper gives an account of some techniques for designing recursive frequency-selective filters which can be applied to data sequences of limited duration which may be nonstationary. The designs are based on the Wiener-Kolmogorov theory of signal extraction which employs a statistical model of the processes generating the data. The statistical model may be regarded as an heuristic device which is designed with a view to ensuring that the resulting signal-extraction filters have certain preconceived properties.