학술논문

Stochastic Modeling and Simulation of Frequency-Correlated Wideband Fading Channels
Document Type
Periodical
Source
IEEE Transactions on Vehicular Technology IEEE Trans. Veh. Technol. Vehicular Technology, IEEE Transactions on. 56(3):1050-1063 May, 2007
Subject
Transportation
Aerospace
Stochastic processes
Wideband
Fading
Narrowband
Wireless communication
Multiaccess communication
Frequency division multiplexing
Rayleigh channels
Closed-form solution
Analytical models
Frequency correlation
Rayleigh fading channels
statistics
stochastic sum-of-sinusoids channel simulators
wideband fading channels
Language
ISSN
0018-9545
1939-9359
Abstract
For the simulation of practical frequency-diversity wireless communication systems, such as frequency-hopping systems, multicarrier code-division multiple-access systems, and orthogonal frequency-division multiplexing systems, it is often desirable to produce multiple Rayleigh fading processes with given frequency correlation properties. In this paper, a novel stochastic wide-sense stationary sum-of-sinusoids channel simulator is proposed to emulate frequency-correlated wideband fading channels, where the frequency correlation properties are controlled by only adjusting the constant phases. Closed-form expressions are provided for all the parameters of the simulation model. This enables us to investigate analytically the overall correlation properties (not only the correlation coefficients) of the simulated processes with respect to both time separation and frequency separation. It is shown that the wideband channel simulator will be reduced to a narrowband Rayleigh fading-channel simulator by removing the frequency selectivity. Furthermore, the COST 207 typical-urban and rural-area channels are applied to evaluate the performance of the resulting wideband and narrowband channel simulators, respectively. The correlation properties of the simulation models approach the desired ones of the underlying reference models as the number of exponential functions tends to infinity, while very good approximations are achieved with the chosen limited number of exponential functions.