Tracking of MIMO Propagation Parameters under Spatio-Temporal Scattering Model

The paper presents novel, computationally efficient methods for tracking radio propagation path parameters from measured tensor-valued MIMO observations. The measurement covariance model is comprised of spatio-temporal scattering plus measurement noise, leading to a shifted Kronecker structure. We derive novel expressions to handle these large (10⁵ times 10⁵) covariance matrices in the extended Kalman filter, used for tracking the dominant propagation paths. Furthermore, we introduce a method to model signals with a given angular distribution, observed with arbitrary antenna arrays. Finally, results for estimating dominant propagation paths from simulated data having an angular (von Mises-Fisher), and temporal (exponential) distributed scattering component are presented, accompanied by a real world example.