Optimization of Beamwidth in Automotive Radars Based on Statistics of Street Geometry

Abstract

In recent years, automotive radars with enhanced cognitive abilities have been used on streets for advanced driver assistance systems to improve driving conditions. Consequently, an ego radar encounters increased automotive radar interference, which causes a significant deterioration in its performance. In this work, we propose exploiting the knowledge of street geometry statistics - specifically the street and vehicular density - to optimize the ego radar’s beamwidth. We model the street geometry as a binomial line process with higher street density at the city centers and wider spatial separation between streets in the sub-urban outer areas. The distribution of vehicles on each street is then modeled as a homogeneous one-dimensional Poisson point process. The resulting binomial line Cox process-based stochastic geometry analysis enables a cognitive automotive radar to optimize its beamwidth to maximize detection performance.