Cluster-Based Scatterer Identification and Characterization in Vehicular Channels

In this paper, we present a new approach for the identification of scattering objects in the delay and Doppler domain. Until now, the identification was done visually based on the power delay profile and video material recorded in the measurement campaigns. We propose to use automatic methods based on the local scattering function (LSF), which brings the Doppler domain into play. The LSF is a multitaper estimate of the two-dimensional (2D) power spectral density in delay and Doppler. Each peak of the LSF is composed of several multipath components (MPCs) coming from the same scattering object. Our approach consists of two steps: (i) detection of the relevant peaks, and (ii) assignment of MPCs to the scattering objects using a clustering algorithm. We use a modified a modified version of the density-based clustering of applications with noise algorithm, where we use the MPC distance. We apply the method to a set of vehicular radio channel measurements and extract the timevarying cluster parameters. The clusters have ellipsoidal shape with their longer axis in the Doppler domain. The first detected cluster presents different properties than the rest of the clusters, being larger, constant in time, and more static in the delay-Doppler plane. By properly identifying only the relevant scattering objects, vehicular channel models, such as the geometry-based stochastic channel model, can be simplified significantly.