Contactless sensing of human physiology
Radio signals reflect from the human body and carry micro-Doppler and range signatures associated with respiration, heartbeat and gross motion. Research over the past decade has demonstrated that carefully processed radar or Wi-Fi signals can estimate breathing rate and, under controlled conditions, heart rate without wearable sensors. 6G sensing extends this capability by providing wider bandwidth, larger antenna arrays and network-level coordination that improve range, angular separation of multiple subjects and robustness to multipath.
Typical processing pipelines extract the phase or displacement time series of a target range bin, apply bandpass filtering matched to physiological frequencies, and estimate rate via Fourier or time-frequency methods. Machine-learning classifiers further map signatures to activity classes or anomaly alerts. Performance depends on orientation, clothing, furniture clutter and the presence of multiple people; multi-static geometries and higher carrier frequencies can mitigate some of these limitations.
Presence, occupancy and fall detection
Even without fine vital-sign resolution, 6G sensing supports reliable presence and occupancy detection in rooms, hallways and outdoor spaces. Change detection on clutter maps or tracking of moving scatterers provides occupancy counts and motion patterns useful for building automation, energy management and security. Fall detection is a higher-stakes application: sudden changes in height and velocity signatures can trigger alerts for elderly care, provided false-alarm rates remain acceptably low through multi-sensor fusion or confirmation logic.
These functions can operate on network infrastructure that already exists for connectivity, reducing the need for dedicated sensors in every room. Latency and reliability requirements, however, may still favor edge processing and local decision logic rather than cloud round-trips.
Gesture and activity recognition
Fine-grained gesture recognition enables contactless human-machine interfaces. High angular and Doppler resolution, combined with temporal models, can distinguish hand motions, pointing gestures or activity sequences. 6G arrays with electronic beam steering can focus energy on a user of interest while suppressing interference from other movers. Training data diversity and domain adaptation remain practical challenges when deploying across different room geometries and user populations.
Privacy, consent and regulatory considerations
Health and biometric sensing is among the most sensitive uses of 6G ISAC. Continuous monitoring of respiration or presence can reveal medical conditions, sleep patterns or occupancy habits. Responsible architectures therefore implement:
- Explicit consent and purpose limitation for any health-related processing.
- On-device or edge processing that avoids exporting raw biometric time series.
- Aggregation and anonymization when data leave the local domain.
- Clear user controls to disable sensing features.
- Auditability of access to sensing results.
Regulatory frameworks for medical devices, biometric data and electronic communications may all apply depending on the claimed functionality and jurisdiction. Systems marketed for clinical decision support face higher evidentiary and certification bars than those offered for wellness or convenience.
System design implications
Supporting health sensing does not require a separate radio. It does require sensing-capable waveforms, sufficient processing budget at the edge, and policy engines that enforce privacy rules. Operators and vendors must decide which capabilities are network-native services, which are application-layer features running on exposed sensing APIs, and how liability and data governance are allocated among the parties in the value chain.
When designed with privacy first, 6G health and biometric sensing can deliver genuine value—safer homes for the elderly, more responsive buildings, and contactless interfaces—while avoiding the surveillance risks that unconstrained environmental monitoring would create.
