Channel Modeling from FMCW Radar

Abstract

An effective modeling of wireless communication channels is crucial for emerging technologies such as autonomous transportation and smart infrastructure, specially to ensure robust connectivity, optimize network performance, and enable adaptive communication under varying environmental conditions. In this thesis, a radar data-driven framework for wireless channel modeling using Frequency Modulated Continuous Wave (FMCW) radar, with a primary focus on path loss estimation in a Vehicle-to-Infrastructure (V2I) single-target scenario, is presented. The proposed methodology is based on signal processing techniques for range and velocity estimation for the detection and separation of target and stationary clutter in the V2I singletarget scenario. Cell-Averaging Constant False Alarm Rate (CA-CFAR) is used to minimize the background and improve the detection of objects in the environment. Furthermore, the clustering is applied to organize clutter patterns and extract relevant features. From the processed data, path loss is calculated separately for both the moving object and the surrounding clutter. These path loss profiles are then fitted to empirical Alpha-Beta (AB) and Alpha-Beta-Gamma (ABG) models capturing overall propagation characteristics. The AB model demonstrates superior fitting performance. The proposed modeling framework characterizes the wireless environment effectively with the data available from radar. This provides systematic foundational methodology for future sensing-based propagation models, used in autonomous systems and smart infrastructure applications.