Description

LOS/NLOS analysis classifies the dominant propagation regime between transmitter and receiver. LOS conditions admit clean Fresnel-zone interpretations and accurate ToF/AoA estimation; NLOS conditions invalidate those models and require either explicit NLOS detection (so they can be discarded) or learned representations that absorb the multipath. In dense indoor venues NLOS dominates, which is precisely why hand-crafted Fresnel models are giving way to deep-learning CSI sensing.

When it's used

  • LOS/NLOS detection as a preprocessing gate
  • Bias correction in indoor positioning
  • Stratifying CSI sensing performance reports
  • Justifying hybrid model + learning architectures

Limitations

  • Boundary between LOS and NLOS is fuzzy in real rooms
  • LOS/NLOS classifiers are brittle across deployments
  • Multipath features that distinguish them overlap with motion features

Source Papers

  • fallani2026_04be — LOS/NLOS in CSI sensing pipeline
  • wu2022_75d3 — LOS dominance in Fresnel-ratio sensing
  • ali2015_d284 — LOS-aware CSI keystroke recognition
  • wang2016_6482 — LOS / NLOS context in WifiU

1 vault paper use this method

Titles and DOIs only — no abstracts, no analyses.

  • A Standard Indoor Spatial Data Model—OGC IndoorGML and Implementation Approaches 2017 DOI ↗