LoS/NLoS Identification in UAV‐Assisted LoRa Networks via Chirp Correlation Features

ABSTRACT

A lightweight signal‐domain framework is proposed to identify LoS/NLoS in UAV‐assisted LoRa networks using the chirp correlation features. At the UAV receiver, two features are extracted, namely, the normalised correlation peak and the width of the −3 dB peak. The results of simulation indicate that the correlation peak has strong discriminative performance, with AUC 0.9 at low‐to‐moderate SNR and almost unity at higher SNR, and the width of the correlation peak provides complementary information. A gated fusion strategy with the consideration of uncertainty also enhances the performance of the strategy in low‐SNR conditions.