Observation-Based Time-Varying MIMO Channel Model

This paper presents a method to model and simulate time-varying narrowband multiple-input-multiple-out (MIMO) channels based on observations from measured data. The data were obtained in a vehicular urban environment, with a fixed transmitter array and a mobile receiver array. The measured channel response matrices were decomposed to isolate the coupling from the transmitter eigenbasis to the received eigenbasis, as in the Weichselberger model. These complex coupling elements have been characterized and seen to comprise directional components that may be Ricean or Rayleigh fading. The Rayleigh fading directional components can be well modeled using the von Mises probability density function, which is parameterized for the time-varying model using the measured data. The model has been validated by comparing the mutual information and eigenstructure autocorrelation characteristics of its output with those of the measured data. The statistical nature of the model means that different realizations can be generated, each representative of the originating data.