Efficient Computation and Covariance Analysis of Geometry-Based Stochastic Channel Models

In this work, we study a family of wireless channel simulation models called geometry-based stochastic channel models (GBSCMs). Compared to more complex ray-tracing simulation models, GBSCMs do not require an extensive characterization of the propagation environment to provide wireless channel realizations with adequate spatial and temporal statistics. The trade-off they achieve between the quality of the simulated channels and the computational complexity makes them popular in standardization bodies. Using the generic formulation of the GBSCMs, we identify a matrix structure that can be used to improve the performance of their implementations. Furthermore, this matrix structure allows us to analyze the spatial covariance of the channel realizations. We provide a way to efficiently compute the spatial covariance matrix in most implementations of GBSCMs. In accordance to widesense stationary and uncorrelated scattering hypotheses, this covariance is static in frequency and does not evolve with user movement.