Performance analysis of AR-model-based linear predictor with Kalman filtering algorithm for wireless communication systems

This paper reports the performance analysis of a proposed auto-regressive (AR) model-based linear predictor algorithm with Kalman filtering (KF). The relationship between the optimum AR order and the channel correlation coefficient is investigated by means of the Akaike Information Criterion (AIC). Through our analysis, 2nd-order differential model based on the AR model-based linear predictor algorithm with KF has the best performance and prediction accuracy. Its performance is about 0.5dB better than a linear predictor algorithm.