A polynomial model for multipath fading channel responses

We present a channel model useful for analyzing the effects of multipath fading in digital radio systems. The frequency response of a fading channel is represented by the function A0 — ωB1 + jωA1, where ω (=2πf) is measured from the center of the channel, and A0, A1,and B1 are variable coefficients that change slowly with time. The model consists of this function, the joint probability density function (pdf) for the three coefficients, and the average number of seconds per heavy-fading month for which this response applies. The model is derived from a large base of multipath fading data, obtained on a 26.4-mile path in Georgia in June 1977. It consists of nearly 25,000 recorded measurements of received power vs frequency in a 26.3-MHz bandwidth at 6 GHz. In this paper, we present the methods of data reduction and statistical analysis used to derive the model; describe some assessments of Us validity; and discuss its limitations, virtues and possible uses. By all available measures, the model is highly accurate. It suffers from a potentially important phase ambiguity that can be resolved only via new, coherent measurements. The existing model should prove very useful in the design and planning of such measurements.