World Sleep Industry ENCYCLOPEDIA

Human Presence Radar

1.3.9

Definition and category boundary

Human presence radar is a contactless sensor that actively transmits radiofrequency signals and analyzes reflections from people and surrounding objects to determine whether a person is present in a defined space and, in some products, where or how that person is moving. It belongs to WSIC 1.3.9. In a bedroom it can provide an occupancy input for lighting, thermal control, ventilation, curtains, nighttime pathways, or care notifications. Some products also expose distance, zone, posture, respiration-like motion, or candidate fall events, but each output requires separate validation and does not follow automatically from the term “human presence.”

The category differs from a passive infrared detector that primarily responds to gross motion, a door contact, pressure mat, camera, ordinary radio transceiver, or radar medical device intended for clinical sleep diagnosis. Millimeter-wave, microwave, ultra-wideband, continuous-wave, frequency-modulated continuous-wave, and impulse radar refer to different bands or waveform architectures; none alone establishes range, spatial resolution, or medical accuracy. Documentation should separate the RF front end, signal processing, presence decision, automation linkage, and any physiological algorithm so that a binary occupancy sensor is not presented as a sleep monitor.

Detection principles and performance measures

Radar derives target information from echo delay, frequency shift, phase, and amplitude. Gross body movement usually produces a strong change. Micromotion such as chest and abdominal movement may help retain a motionless person, but it is more sensitive to mounting stability, breathing amplitude, posture, bedding, and environmental vibration. Different architectures expose different combinations of presence, range, velocity, angle, and zone. Products in the same band can still differ because of antenna design, bandwidth, waveform, power, sampling, processing chain, and algorithm. Any claimed “accuracy” needs a stated measurand, reference, distance, posture, number of people, and statistical definition.

Occupancy performance should at least separate probability of detection, false absence, false presence, entry response time, stationary retention, vacancy delay, valid field, and suppression outside the boundary. Range resolution is not angular resolution, and reliable detection of one person does not establish separation of two bed partners. Greater penetration is not always desirable: detection through a wall or across zones can destroy room attribution. Presence, candidate fall detection, respiratory rate, heart rate, sleep versus wake, and sleep staging are distinct tasks whose signal quality, reference methods, and validation populations cannot be exchanged.

Bedroom environment and installation

Placement should follow the target bed, room boundary, antenna field of view, permitted range, and manufacturer instructions, with height, tilt, azimuth, and obstruction recorded. A metal bed frame, mirror, corner, or other large reflector can create multipath. Thick bedding, a canopy, cabinetry, and body orientation can attenuate or alter micromotion echoes. Fans, moving curtains, pets, robot cleaners, people in an adjacent room, vibrating equipment, and moving doors can all become interference sources. Wood, glass, and lightweight partitions may pass some signal, but a material name alone is insufficient to predict attenuation or false detection.

Each zone should be commissioned after the furniture, bed, bedding, and normal activity pattern are in place, not only in an empty room. Boundaries need to cover actual routes such as the doorway, bedside, and bed surface while avoiding walls or spaces that should not be monitored. Range gates, sensitivity, stationary thresholds, and vacancy delays interact. Higher sensitivity may reduce false absence while increasing multipath and nonhuman motion. A loosened bracket, wall vibration, a changed angle after cleaning, or rearranged furniture can invalidate the original setting.

Multiple people, stillness, and abnormal conditions

Very deep stillness, a body facing away, prone posture, thick covers, shallow breathing, or a position near the field edge may yield insufficient echo. Overlapping people, bed partners, children, pets, and a caregiver entering the room can mix targets. An algorithm that maintains a track does not prove continuity of identity or correct attribution to an individual. Bedding that continues moving after a person leaves, a bedside fan, or activity across a wall can delay a vacancy decision. Documentation should state the maximum target count, occlusion behavior, what happens after track loss, and which outputs are disabled in a multi-person scene.

“No person detected” does not prove an empty room or a medical emergency, and “periodic micromotion detected” does not confirm normal breathing or safety. Lighting and comfort automation needs manual override, reasonable delays, and a conservative fault state so that necessary light, ventilation, or thermal control is not removed by an error. Care notifications for an older adult, child, or person unable to call for help should combine other sensing, human confirmation, and an explicit response process. A consumer presence radar must not be the sole life-safety monitor.

Automation, faults, and cybersecurity

Radar output should expose understandable states such as occupied, vacant, zone, confidence, or fault rather than mapping an opaque algorithm change directly to a hazardous action. Automation should define entry and vacancy delays, hysteresis, minimum run times, schedules, manual override, and power-recovery behavior. Local limits and interlocks for heating, ventilation, lighting, locks, and motorized devices remain at the actuator. A sensor stuck occupied or vacant, a stale timestamp, gateway loss, congested networking, and software restart all require testable degraded behavior.

A connected product should use unique device identity, controlled commissioning, authenticated and encrypted communication, least privilege, secure updating, event records, and credential deletion. Presence, range, nighttime movement count, bed dwell, and respiration-like waveforms can reveal occupancy, routines, and health information even when no image is produced. The product should say whether raw echoes, features, and events are processed locally or in the cloud, how long they are retained, who can access them, and how they can be exported and deleted. Installer, household, hotel-operator, and care-organization roles need separation, and remote tuning must not defeat local safety boundaries.

Radiofrequency, electrical, and regulatory fit

Human presence radar is an active RF device and should not be marketed as producing “zero radiation.” Compliance also cannot be inferred solely from low power or short range. Frequency band, occupied bandwidth, transmit power, power spectral density, antenna gain, duty cycle, and unwanted emissions must meet the radio rules applicable where the product is sold and used, evaluated with the final enclosure, antenna, firmware, and operating modes. Human RF exposure assessment should account for frequency, separation, direction, simultaneous transmitters, and installation. A report for one chip or module cannot automatically be extended to every finished-product configuration.

Power supply, temperature rise, overvoltage, protection against electric shock, electromagnetic compatibility, enclosure, and environmental conditions also matter. An in-wall or mains-powered product, or one connected to a home and building control bus, should be installed by people qualified under local requirements. An extra-low-voltage module still cannot be placed casually with mains conductors. Radio authorization addresses spectrum and related technical requirements; it is not certification of the presence algorithm, cybersecurity, medical performance, or the complete automation system. Diagnostic, alarm, or monitoring claims may also trigger product requirements different from those for an ordinary automation sensor.

Commissioning, maintenance, and claim limits

Commissioning under the real furniture and user conditions should cover entry, exit, prolonged stillness, turning, getting out of bed, door movement, different sleep postures, and bedding. It should add challenges from multiple people, pets, fans, curtains, adjacent-room activity, and network loss. Records should include true detections, false absence, false presence, response and vacancy delay, cross-zone events, and the resulting automation rather than only a live visualization. If respiration, heart rate, falls, sleep/wake, or sleep stages are reported, the device needs a reference method suited to that intended purpose and independent validation across representative people, postures, conditions, and home environments.

Maintenance includes checking mounting angle, removing obstructions, reviewing zone boundaries, events, and privileges, installing validated updates, and retesting after furniture movement, bed replacement, addition of a pet, algorithm upgrade, or gateway change. Research shows that particular radar systems and algorithms can estimate respiratory, cardiac, or sleep-related variables under defined conditions, but results cannot be transferred directly across models, bands, placements, and populations. Ordinary presence detection cannot diagnose sleep apnea, insomnia, or another disorder and does not replace polysomnography, a validated medical device, or professional assessment.