Adaptive noise cancellation with nonlinear post processing

A study of adaptive noise cancellation combined with nonlinear postprocessing for the purpose of signal detection is presented. First, a model for the transient signal to be detected is provided. This signal is subject to interference and noise from multiple sources. The properties of these noise and interference components are given. An adaptive filter in the predictor configuration whose coefficients are updated by the least mean square algorithm is used in the first step of the noise cancellation process. The predictor output of the adaptive filter contains the predictable components of the signal plus noise input. The signal to be detected, due to its sudden onset which is in contrast with the steady presence of the noise, is difficult to predict by the filter, and therefore is seen in the error output of the adaptive filter. The error signal is then passed through a Kaiser-Teager (1990, 1980) energy operator which significantly enhances the transient signal to be detected