MmWave NLOS Sensing under Rough Relay Surface: Challenges and Solutions

Non-Line-of-Sight (NLOS) perception is essential for preventing potential collisions and finds applications in areas such as autonomous driving, target localization, and human recognition in urban environments. Millimeter-wave (mmWave) radar distinguishes itself by robust NLOS sensing abilities. However, in urban environments, the reflective properties of relay surfaces have become complex. Diverse construction materials, intricate architectural designs, and dynamic environmental variations introduce complexity, violating the conventional assumption of smooth, deterministic geometries and degrading NLOS perception performance. To improve mmWave NLOS sensing under these realistic conditions, we investigate the characteristics of rough relay surfaces and develop a sensing framework to manage multipath interference. Our approach specifically focuses on three specific challenges often overlooked in current research: precise detection under small-scale rough relay surfaces, high-resolution imaging under large-scale relay surfaces, and multipath scattering classification under uncertain surfaces. We establish scattering signal models under different roughness conditions and develop effective NLOS sensing systems based on the distribution characteristics of targets behind different rough surfaces, achieving accurate and reliable sensing.