A Stochastic Model of the Time-Variant MIMO Channel Based on Experimental Observations

This paper proposes a stochastic multipath model that is useful for generating multiple-input-multiple-output (MIMO) channel matrices in time-variant environments. The multipath model is developed by first extracting the relevant multipath cluster characteristics from measured indoor channel data and subsequently capturing these characteristics in an autoregressive stochastic model. This model is then used to generate channel matrices whose space-time characteristics closely match those of realistic scenarios, particularly when birth and death of multipath clusters are included in the stochastic representation. Computational examples reveal the applicability and the accuracy of the approach. While the current implementation is based on data taken assuming that propagation is confined to the horizontal plane, this paper also discusses the extension of the model to describe 3-D propagation, enabling its application to a wide range of physical scenarios and antenna characteristics.