TABLE OF CONTENTS In generating signals similar to those encountered in mobile communications, it is practical to compute signal samples from the scenario model by moving the receiving antenna amongst discrete sources. In simulations, calculations of any metric over a finite sample size will provide only an estimate of the ensemble average value. In this section, basic considerations for computer simulations of the short term fading channel are discussed. Two aspects of simulation are: the generation of discrete channel samples, that may be mutually independent and have the correct ensemble averaged statistics; and the generation of time (i.e. space ) series which are samples of continuous channel behaviour, and these may comprise closely spaced, correlated samples, which should also have the correct ensemble averaged statistics.
Two dimensional scenarios are the most used. In order to produce accurate multipath statistical distributions, such as Rayleigh distributed envelopes, a minimum of about ten equiamplitude and random phase sources should be received at any one time. The sources need not have the same amplitude for simulation purposes, the main point being that the real and imaginary parts approach Gaussian distributions. More sources than this may be required to achieve uncorrelated samples in long record lengths (see Section 7.1.4). In this section, the uniform scenario is of main interest.
The uniform, or Clarke, scenario is simulated using a ring of discrete sources with the receiver at the nominal centre of the ring. The sources can be equal amplitude, equal excess delay, and random phase. This...
