Prediction, smoothing and interpolation in adaptive diversity reception

Adaptive channel estimators based on linear prediction, or smoothing and interpolation are analyzed and compared in a flat fading channel. Data modulation is removed by using decision feedback before the estimator based on prediction. In the estimator based on smoothing and interpolation known training symbols are exploited. It is shown how the estimators can adaptively be "matched" to the Doppler power spectrum of the channel by using the well-known algorithms such as the LMS and RLS algorithms. The method has potential applications in mobile communications, e.g. in spread-spectrum RAKE receivers and in other diversity receivers.<
>