Study of Autoregressive Model Order for Fading Radio Channel Prediction

In the close-loop adaptive transmission mobile wireless systems, information about channel state (CSI) at the receiver is fed back to the transmitter. To enable an efficient adaptive transmission, CSI has to be predicted several symbols ahead to compensate a delay in the feedback loop and adaptation rate limitations. Autoregressive model based linear predictor is addressed in the paper, particularly an impact of predictor order on prediction accuracy for various fading scenarios is investigated. The results show that the memory length is the important factor that affects CSI prediction accuracy especially for fast fading channels.