Microstatistic recursive least squares

A piecewise-linear signal model based on amplitude threshold decomposition underlies the idea of microstatistic nonlinear signal characterization. This model is easily incorporated into many existing linear signal processing techniques and leads to algorithms that are robust (e.g., tolerant of non-Gaussian noise) and that are applicable to nonlinear signal processing problems. In this contribution, we incorporate signal amplitude threshold decomposition in the recursive least squares (RLS) algorithm. The microstatistic RLS algorithm can be used to process nonlinear signals and should find application in areas such as communications, geophysical signal processing, and biomedical signal analysis, among others.<
>