The Common Thread in Optimal Adaptive Non Recursive Filters.
Abstract
Filters are applied to time sequences to extract desired signals from random noise and from interfering signals. The specific form of the filter is related to apriori knowledge of the statistics of the desired and the undesired signals. Classic filter design entails two distinct stages. The first is to employ some monitoring and smoothing scheme which will lead to estimates of the signal statistics such as the signal and noise covariance functions or power spectrums. The second stage is to formulate the filter response in terms of these estimated statistics. In this paper we review a class of filters for which the two stages just described are performed simultaneously and iteratively. The filter coefficients are changed by a recursive algorithm which corrects the filter response during the processing of the input data. The capability to modify the filter response during operation makes it possible to track and to filter signals with slowly changing statistics. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 04, 1981
- Accession Number
- ADA122850
Entities
People
- Frederick J. Harris
Organizations
- San Diego State University