Modeling of radio channels with rayleigh fading and non-isotropic spectrum of Doppler spread

Wireless telecommunication systems should be designed for operation over fading radio channels under great variations of environment conditions. An accurate simulation of those channels is required during design and testing of both algorithms and modern radio communication system hardware. This work is concerned on autoregressive channel model of moving average with Rayleigh fading and non-isotropic spectrum of Doppler spread. Proposed model is based on digital autoregressive moving average (ARMA) filter. Simulation uses source filtered complicated white Gaussian noise to produce (approximately) time selective channel process with required non-isotropic spectrum of Doppler spread. This Doppler spread of non-isotropic scattering occurs due to real antenna directional parameters and some reflections along propagation path. We apply infinite impulse response (IIR) low-pass digital filter for approximation of required spectral characteristics and use von Mises distribution for angles of arrival description. Spectral power density of reference model fading process if realized in the form of its approximation by the square of digital filter transmission function with complex coefficients. We focused on estimation of spectrum-shaping filter order and suitable initial approximation for the following numerical solution of Chebyshev approximation problem with reduced number of arithmetic operations produced by this filter.