摘要 |
Loop gain normalization is employed in adaptive filters to control weighting of the filter characteristic updates in order to converge properly to a desired filter characteristic. Filter instability is avoided during intervals that transient or other rapidly pulsating signals are received or inputted by normalizing the update gain with a representation of a so-called fast attack estimate of a prescribed characteristic of the input signal. In one embodiment the fast attack estimate is the maximum of a plurality of power estimates generated from a corresponding plurality of subsets of amplitude samples of the received signal. In another embodiment, the maximum of the magnitude representations of the input signal samples is modified to represent the fast attack power estimate of the received signal.
|