Improved Self Tuned Linear Predictor

An improved self tuned linear predictor used in the separation of interference signals is presented in this work. From the classic structure of a Wiener adaptive filter we have adapted and included a Modified NLMS algorithm instead of the conventional NLMS first proposed by Widrow. Working in real time, this predictor allows the use of a FIR Wiener filter, of smaller order, for equal speed of processing. The predictor here presented has been successfully tested under different conditions of operation, with digital simulation. Compared with identical predictors using the NLMS algorithm, it showed a higher fidelity in the predicted signal and a greater speed in the process of prediction. These characteristics can make attractive their use in real time applications.