Decision-directed tracking of fading channels using linear prediction of the fading envelope

A decision-directed adaptive channel-tracking scheme can fail in a fading environment. We propose a method for overcoming this problem using a prediction of the complex fading envelope, which is modeled as a bandlimited random process. Making use of the prolate spheroidal wavefunctions, we solve the bandlimited linear prediction problem for the special case in which the process spectral density is flat within the band limits. How these findings may be applied to the channel tracking problem, as well as other more complex fading compensation measures, is also discussed.