Description
Binary detection of whether a person is present in a monitored area, including stationary occupants. Presence detection is the harder counterpart of motion-detection because it must distinguish a still-but-breathing human from a truly empty room — the small (mm-scale) chest-wall motion of breathing is the only signal available. Robust presence detection is the prerequisite for trustworthy occupancy-estimation and for any energy-management system that must turn off services in genuinely empty spaces.
Why it's hard
- Stationary humans are nearly invisible to motion-thresholding pipelines.
- Breathing-band signals are weak and easily masked by environmental periodicities (HVAC, fans).
- Empty-room CSI baselines drift, producing slow-rising false positives.
- Distinguishing a stationary human from a sleeping pet or a moving curtain is non-trivial.
- Multi-occupant presence requires joint detection that aggregates breathing-band evidence.
Common approaches
- CSI breathing-band spectral analysis (0.1–0.5 Hz) for stationary-human detection.
- Combined motion + breathing detectors for staged inference.
- Self-supervised baselines learned per-room from confirmed-empty intervals.
- Multi-link diversity (multiple AP-client pairs) to gain spatial coverage.