What is radar fall detection?
Radar fall detection uses short-range radio waves to monitor a room and detect when someone falls. No cameras, no microphones, nothing for the person to wear. A small sensor mounted on the wall continuously emits low-power 60GHz radio waves (also called millimetre waves or mmWave) that bounce off people and objects in the room. By analysing how those reflections change over time, the system can tell what is happening: someone walking, sitting down, lying in bed, or falling to the floor.
If you have driven a modern car with automatic emergency braking, you have already trusted this technology with your life. The same physics (FMCW radar) is what prevents your car from rear-ending the vehicle in front. It is also used in airport security scanners, hospital patient monitoring, and industrial safety systems.
The difference in elder care is that the radar is tuned for indoor, room-scale detection. Instead of tracking vehicles at 100 metres, it tracks human movement at 1–8 metres. Instead of triggering brakes, it sends an alert to a family member's phone.
The science: how FMCW radar works (simply)
FMCW stands for Frequency-Modulated Continuous Wave. Here is what that means in plain terms:
- The sensor sends out a continuous radio signal, not a pulse, but a smooth wave that sweeps across a range of frequencies. Think of it as a constant, invisible hum of radio energy.
- The signal bounces off everything in the room: walls, furniture, and people. Each reflection returns to the sensor slightly altered.
- The sensor compares what it sent with what came back. The tiny differences in frequency and timing reveal two things: how far away the reflecting object is, and how fast it is moving.
- Software analyses these reflections thousands of times per second, building a real-time "motion profile" of the room. It knows where movement is happening, how fast, and in what direction.
Crucially, 60GHz radar does not produce an image. There is no picture to see, no video to record, no face to recognise. The data is a stream of numbers representing distance, velocity, and signal strength. A person looks like a cluster of moving reflection points, not a recognisable human figure.
How radar distinguishes a fall from normal activity
This is the clever part. Every movement creates a distinctive signature in the radar data. The system's algorithms (typically machine learning models trained on thousands of real and simulated falls) look for specific patterns:
A fall has a distinctive radar signature
- Rapid downward velocity. A person falling moves towards the floor much faster than someone sitting down or bending over. The vertical velocity component is a strong differentiator.
- Height change. The main reflection cluster drops from standing/sitting height (1.0-1.7m) to floor level (0-0.3m) within 1-2 seconds.
- Impact signature. The moment of hitting the floor creates a brief, distinctive pattern in the reflected signal: a sudden deceleration followed by micro-oscillations.
- Post-fall immobility. After a fall, the reflection pattern changes dramatically. Instead of the varied, dynamic reflections of a moving person, the sensor sees the subtle, rhythmic pattern of someone lying still, perhaps just breathing. This immobility confirmation is what separates a confirmed fall from a controlled movement to the floor.
What does NOT look like a fall
- Sitting down: slower descent, controlled velocity, height drops to chair/sofa level (0.4-0.6m) not floor level
- Lying down in bed: horizontal movement towards the bed, gradual height reduction, located at bed height not floor level
- Bending over: brief height reduction followed by return to standing height. No impact signature. No sustained low position.
- A child or pet moving: different reflection profile (smaller cross-section, different height range, different movement patterns)
Modern algorithms combine all of these features (velocity, height, impact, location, and post-event behaviour) to make a determination. The best systems achieve over 95% sensitivity (catching real falls) with less than 1% false-positive rate in peer-reviewed studies.
Advantages over accelerometer-based (wearable) detection
Smartwatches and pendants with automatic fall detection use accelerometers, tiny sensors that measure the acceleration forces acting on the device. When the device (and therefore the person's wrist or chest) experiences a sudden acceleration consistent with a fall, it triggers an alert.
Accelerometers work, but they have inherent limitations that radar does not share:
| Factor | Accelerometer (wearable) | Radar (ambient) |
|---|---|---|
| Must be worn | Yes (useless otherwise) | No (works passively) |
| Needs charging | Yes, daily for smartwatches | No, mains powered |
| Detects slow falls/collapses | Poorly: low-acceleration events are missed | Well: height change and immobility are key signals |
| False positives | Common (arm gestures, dropping items, bumping into things) | Rare (full-body movement analysis is more robust) |
| Works in bathroom | Only with waterproof models worn in the shower | Yes, radio waves are unaffected by moisture |
| Works during sleep | Only if worn in bed | Yes, always on |
| Detects post-fall immobility | Limited: detects lack of movement but not position | Yes, confirms person is on the floor |
The fundamental difference is this: an accelerometer measures what the device is doing. Radar measures what the person is doing. The device can be left on the nightstand, forgotten in a coat pocket, or thrown in a drawer. The person is always in the room.
Privacy: what radar can and cannot see
This is worth being precise about, because the data looks nothing like what most people expect.
Radar CANNOT:
- Produce images or video of any kind
- Identify who someone is (no facial recognition, no biometrics)
- Record audio or conversations
- See through walls into other rooms (60GHz signals are absorbed by standard building materials)
- Determine what someone is wearing, doing with their hands, or looking at
Radar CAN:
- Detect that someone is present in the room
- Track their general position and movement speed
- Determine their approximate posture (standing, sitting, lying)
- Detect a fall event
- Measure breathing rate at close range (through chest wall movement)
- Track activity levels over time (how much movement throughout the day)
This is what makes radar fundamentally different from cameras. A camera captures identity, context, behaviour, and appearance. Radar captures presence, movement, and position. You cannot reconstruct a radar data stream into anything resembling an image of a person. The physics does not allow it.
For families, this means genuine privacy. Your parent can go about their day, getting dressed, using the bathroom, watching television, with the certainty that nobody can see them. The system knows whether they are moving normally or not, and nothing else. For a broader look at privacy-respecting monitoring options, see Camera-Free Home Monitoring: How to Watch Over Parents Without Invading Privacy.
Where radar is already trusted
If 60GHz radar in the home sounds unfamiliar, consider where it is already used:
- Automotive safety: every major car manufacturer uses FMCW radar for adaptive cruise control, automatic emergency braking, blind-spot detection, and parking assistance. If you have driven a car made after 2020, you have used this technology.
- Airport security: millimetre-wave body scanners (the ones you walk through with arms raised) use the same frequency band to detect concealed objects without physical searches.
- Hospital patient monitoring: contactless vital sign monitoring using radar is used in sleep labs and intensive care units to track breathing and heart rate without attaching sensors to the patient.
- Smart buildings: occupancy detection for lighting and HVAC systems increasingly uses 60GHz radar instead of PIR sensors for greater accuracy.
- Consumer devices: Google's Pixel phones have used radar (Project Soli) for gesture control since 2019. The Nest Hub uses radar for sleep tracking.
The technology is mature and already deployed at enormous scale. Applying it to elder care fall detection is a relatively straightforward extension of capabilities proven in more demanding environments.
Health and safety: is 60GHz radar safe?
Yes. The power levels used in indoor radar sensors are a fraction of those emitted by a mobile phone or Wi-Fi router. The International Commission on Non-Ionizing Radiation Protection (ICNIRP) has established safety guidelines for 60GHz exposure, and consumer radar sensors operate well below these limits, typically by a factor of 100 or more.
At 60GHz, radio waves are absorbed by the outer layer of skin (the first 0.5mm) and do not penetrate into the body. They cannot cause tissue heating at the power levels used in consumer devices. For comparison, a few hours of sunlight deposits more energy into your body than a lifetime of exposure to a 60GHz radar sensor.
What this means for elderly care
Radar-based fall detection solves the two problems that have plagued the industry for decades: compliance and coverage.
Compliance is solved because there is nothing to comply with. The sensor is on the wall. It works whether the person knows about it or not (with their consent, of course), whether they remember it or not, and whether they are having a good day or a bad day.
Coverage is solved because radio waves do not care about lighting, time of day, or moisture. The bathroom at 3am (statistically the most dangerous combination of place and time for an elderly person) is covered just as well as the living room at noon.
For families, this means that if your parent falls, you will know about it within seconds, regardless of the circumstances. For care providers, it means scalable, consistent monitoring that does not depend on client behaviour.
For a full comparison of how radar stacks up against pendants, smartwatches, and cameras, see our guide: Fall Detection for Elderly: Every Option Compared (2026). For detailed product-by-product comparisons, see Fall Detection Devices Compared: Pendants vs Watches vs Sensors.