On the problems of Monte Carlo method based simulation models for mobile radio channels

Propagation models for mobile radio channels are often derived by using Gaussian noise processes with given Doppler power spectral density shapes. Such stochastic processes can efficiently be realized by employing Rice's sum of sinusoids. In the meantime, the Monte Carlo principle is widespread in use for designing the parameters of the sinusoids. This paper investigates the (higher order) statistical properties of Monte Carlo method based simulation models for Rice and Rayleigh channels. It will be shown analytically and experimentally that a single realization of the output sequence of the simulation model does in general not contain the aspired statistical information. Moreover, an alternative parameter computation method will be discussed, which avoids the disadvantages of the Monte Carlo method.