Model-aware UAV trajectory planning for efficient radio environment mapping

Accurate Radio Environment Mapping (REM) is essential for wireless network optimization and performance analysis. However, conventional data acquisition and REM construction methods face practical challenges, particularly in complex or inaccessible urban areas. Model-driven approaches such as ray tracing (RT) offer a promising alternative but are often limited by unknown electromagnetic (EM) parameters that reduce their accuracy. To overcome this, we propose a model-aware UAV trajectory planning framework for efficient REM generation in challenging environments using limited measurements. The problem is formulated as a joint UAV path planning and REM generation as an inverse modeling problem to collect informative data for inferring unknown EM parameters, thereby improving RT-based REM reconstruction. Simulations in an urban setting demonstrate that the proposed method improves REM estimation accuracy by up to 40% compared with baseline approaches. This work presents the first model-aware UAV trajectory optimization framework for EM parameter learning, enabling accurate and data-efficient REM construction in inaccessible regions.