Simulation of doubly selective fading channels using autoregressive modeling and optimum receiver design

Abstract

Autoregressive (AR) modeling is a well known scheme for signal and system modeling. In this paper, we investigate using AR models to generate doubly selective fading with correlated and uncorrelated paths. We briefly consider an autoregressive moving average (ARMA) extension. AR models are shown to closely approximate channel correlation functions for reasonable model orders. When we use the AR/ARMA-generated channel taps in the receiver to replace channel estimation, the sacrifice in receiver performance is shown to be insignificant. The AR/ARMA approaches can be used to reduce complexity at the receiver side by using low-complexity filters.