Adaptive general parameter extension to FIR predictors

In this paper, a general parameter (GP) extension to polynomial and sinusoidal FIR predictors is proposed for extended sinusoidal and Rayleigh fading signal prediction. With a single adaptive general parameter, it is shown possible to extend prediction capabilities of polynomial and sinusoidal FIR predictors beyond polynomial input signals, or beyond the nominal design frequencies. This allows for more accurate prediction of input signals with unknown time varying statistics. The form of adaptation proposed can be regarded as the simplest possible adaptive filtering, and is shown to yield improved filtering performance at low computational cost. It is demonstrated that the proposed method provides for improved prediction of sinusoidal and Rayleigh distributed signals. Here, Rayleigh distributed signal models the received transmission power in a mobile communications system, and the sinusoid is a model of the power line frequency.