An empirical multi-wall NLOS ranging model for Wi-Fi RTT indoor positioning

This paper proposes a novel empirical non-line-of-sight (NLOS) ranging model, called a multi-wall model, for Wi-Fi round-trip-time (RTT) indoor positioning. It is purposely designed to account for NLOS ranging biases due to through-wall signal propagation in complex indoor environments. Also, a low-order polynomial term is added to capture unmodeled NLOS effects. The model takes a 2D geometry floor plan as prior input and learns its parameters from a few on-site training data. It is evaluated by experiments using a large public RTT dataset and our self-collected data and compared to the naive LOS ranging, the polynomial-based empirical model, and a data-driven model. Results show that our model has superior NLOS-ranging performance. At last, we show the application of our model in fingerprinting-based RTT positioning with experiments.