Validation of minimum-energy band-limited prediction using vehicular channel measurements

We demonstrate that minimum-energy (ME) band-limited prediction shows the same robust performance for vehicular channel measurements as well as for the numeric Clarke channel model. By contrast, channel prediction based on sinusoidal modelling presented by Chen et al., 2007, shows poor performance for a small percentage of measured channel realizations. This increases the mean square error dramatically, hence outlier removal is required. The ME band-limited predictor introduced by Zemen et al., 2007, is based on a subspace spanned by time-concentrated and band-limited sequences. The time-concentration of these sequences is matched to the length of the observation interval and the band-limitation is determined by the support of the Doppler power spectral density of the fading process. The low-complexity time-variant flat-fading channel predictor dynamically selects a predefined subspace from a small set such that the prediction error is minimized. We validate the ME band-limited predictor using channel measurements from an alpine region. The predictor performance with measured channels is comparable to the one obtained with Clarke's channel model for non line-of-sight situations. For line-of-sight situations the performance is better than for Clarke's model. We present results in terms of mean square error averaged over all measured snapshots.