Finite sample estimates for mobile channels

The narrowband channel of mobile communications has a Rayleigh-like distribution for the envelope in space (or time, for a moving terminal in a static scenario) and also in frequency. As the channel signalling bandwidth increases to become an appreciable fraction of the channel coherence bandwidth, the variation in channel gain from the multipath reduces because of the integration over frequencies which are partially correlated. This is the advantage of wideband channels for communications. Analogously, averaging of the channel gain in space results in a decreased gain variation than for a single point in space (cf the advantage of space diversity). For statistical calculations of radio coverage, it is important to characterize the accuracy of the channel gain estimates. Also, estimating the next moment, the correlation coefficient to a known accuracy is also important for antenna or frequency diversity design. This paper gives the averaging results and finite-sample interpretation of spatial and frequency integration for the mean power (channel gain) and the correlation coefficient.