Multi-Cluster Frequency-Angle Diffuse Scattering Model and Its Estimation

For the development and deployment of upcoming 6G systems, propagation channel measurements in new scenarios, new frequency bands, and for new antenna arrangements will be required. The diffuse multipath component (DMC) contains a non-negligible portion of the channel impulse response, and thus must be considered in channel evaluations and modeling. Previous work observed that multiple DMC clusters exist, which are associated with specular multipath component (MPC) clusters, yet the widely used Kronecker assumption for the delay/angle structure of these DMC clusters rarely holds in reality. In this paper, we propose an estimation algorithm for a parametric multi-cluster DMC model that more accurately models the connection between delay and angular parameters for each cluster, assuming that the Kronecker model is only valid within each cluster instead of for the whole channel. This avoids creating ghost modes that would reduce the DMC covariance matrix estimation accuracy. Our proposed algorithm is incorporated into the framework of the RiMax algorithm, an iterative maximum-likelihood estimation scheme. The effectiveness and correctness of the proposed algorithm are verified on synthetic channels, since they have a known ground truth. Simulations demonstrate improvements not only in the accuracy of the DMC, but also the associated specular MPC compared to the state-of-the-art uni-modal DMC model. Furthermore, the application of this algorithm to real-world sub-THz channel measurements is demonstrated.