A method for adaptive filtering an input sequence of sample values and for the adaptive linear combination of sampled values and adaptive linear combiner

摘要

The method involves an adaptive filtering of an input sequence of sample values of a signal. An error value is formed from a respective output value and a corresponding reference value, as a measure for the deviation between those values. New coefficients for a respective next filter step are calculated from a filter coefficient of a current filter step. The inverse of a correlation matrix is used when calculating the new coefficients, which indicates, through its elements, the correlation between the sample values of the input vectors of a section in the respectively preceding filter step. A new inverse of a new correlation matrix is calculated from the inverse for the coefficient determination in the respectively next filter step. The section has a constant length during at least two successive filter steps.