Statistical Model of Spatially Correlated Shadow-fading Patterns in Wireless Systems

We discuss a statistical model to generate correlated shadow-fading patterns for wireless systems in the absence of detailed propagation and landscape information. The current model of shadow-fading postulates a log-normal marginal distribution for the fading values, and does not address correlations. Subsequent introduction of correlations via autocorrelation models for individual mobiles results in anomalous effects that depend on traffic density and mobility. Our approach involves generating a pre-computed set of fading values with the right marginal distributions and spatial correlations. Motivated from statistical physics, the correlations are introduced in terms of “interaction” parameters, which can be computed from local measurements. The model is efficiently implemented using standard linear-algebra methods and is amenable to a statistical mechanics treatment. Numerical results show that the patterns produced are sufficiently clustered and appear reasonable to visual inspection.