1: \begin{abstract}

2: In this work, we study a family of wireless channel simulation models called geometry-based stochastic channel models (GBSCMs).

3: 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.

4: The trade-off they achieve between the quality of the simulated channels and the computational complexity makes them popular in standardization bodies.

5: Using the generic formulation of the GBSCMs, we identify a matrix structure that can be used to improve the performance of their implementations.

6: Furthermore, this matrix structure allows us to analyze the spatial covariance of the channel realizations.

7: In accordance to wide-sense stationary and uncorrelated scattering hypotheses, this covariance is static in frequency and does not evolve with user movement.

8: Finally, we show that convergence may be slow when using GBSCMs to evaluate the performance of covariance-based algorithms, and propose a solution to alleviate this issue.

9: \end{abstract}