Optimal Placement of a Moving Sensor for Passive Localization in a Real NLoS Environment

The site-specific non-line-of-sight (NLoS) conditions and unpredictable transmission signals in urban areas complicate localization efforts. Radio-frequency fingerprinting (RFF) addresses this challenge by building a database of signal characteristics at various locations. However, transition from indoor to outdoor environments is difficult due to the vast physical area and inaccessible sites. In this paper, we propose an RFF-based passive localization approach enhanced by ray-tracing simulation. This simulation utilizes real geographic data, including all buildings and terrains, without making assumptions about NLoS error statistics. To improve localization accuracy, we investigate the optimal placement of a moving sensor by minimizing the average mean squared error (MSE) within a confidence interval. Simulation results indicate that RFF-assisted passive localization is effective in real-world scenarios, and the optimal placement of the moving sensor significantly enhances localization accuracy.