A Novel Approach to the Statistical Modeling of Wireline Channels

Abstract: We report here that channel power gain and Root-Mean-Square Delay Spread (RMS-DS) in Low/Medium Voltage power line channels are negatively correlated lognormal random variables. Further analysis of other wireline channels allows us to report a strong similarity between some properties observed in power line channels and the ones observed in other wireline channels, e.g. coaxial cables and phone lines. For example, it is here reported that channel power gain and logarithm of the RMS-DS in DSL links are \textit{linearly} correlated random variables. Exploiting these results, we here propose a statistical wireline channel model where tap amplitudes and delays are generated in order to reflect these physical properties. Although wireline channels are considered deterministic as their impulse response can be readily calculated once the link topology is known, a statistical wireline channel model is useful because the variability of link topologies and wiring practices give rise to a stochastic aspect of wireline communications that has not been well characterized in the literature. Finally, we also point out that alternative channel models that normalize impulse responses to a common (often unitary) power gain may be misleading when assessing the performance of equalization schemes since this normalization artificially removes the correlation between channel power gain and RMS-DS and, thus, Inter-Symbol Interference (ISI).