Optimal mean-square prediction of the mobile-radio fading envelope

Long-range prediction of the mobile-radio fading envelope is an enabling technology for many fading compensation approaches. Because the fading envelope is well modeled as a bandlimited process, it has special predictability properties. In this paper, we find a linear predictor that is optimal in the mean-square sense when the predictor impulse response is energy constrained. This solution may be used to determine the minimum mean squared error of a prediction based on past values that are corrupted with estimation errors.