Skip to Main Content
While the frequency dependence of the wireless channel may be negligible for narrow to wideband signals, it has been shown that modeling this dependence for bandwidths in excess of 2 GHz improves channel reconstruction up to 40%. Yet to our knowledge, only Molisch have done so for the ultrawideband channel. Their benchmark frequency model however represents the average dependence over the collection of multipath arrivals in the channel rather than that of individual arrivals. Building on the geometric theory of diffraction, we propose a novel stochastic frequency model for individual arrivals according to the propagation events on their paths between the transmitter and receiver. We extract the model parameters from an extensive measurement campaign of 4000 channel frequency sweeps in four separate buildings combined with raytracing simulations, and show that ours fits the gathered data more closely than the benchmark model.