The phrase “mmWave radar can detect still people” is easy to misunderstand. It can sound as if a radar naturally knows there is a person sitting still in the room.
That is not the right model.
Radar receives electromagnetic echoes. It observes distance, angle, reflection strength, velocity, phase, and how those values change across frames. Whether those changes mean “a human is present” is an algorithmic interpretation.
The first model is: mmWave presence radar does not detect the concept of a human directly. It observes echo energy, velocity, phase, range bins, angle bins, and multi-frame changes, then classifies some patterns as human presence.
Transmit mmWave signal
-> Objects reflect electromagnetic waves
-> Receiver collects echoes
-> Estimate distance, velocity, angle, and phase
-> Use background modeling and micro-motion features
-> Infer human presence
What It Transmits
mmWave radar transmits electromagnetic waves in or near millimeter-wave bands. Common modules may work around 24 GHz, 60 GHz, 77 GHz, or 79 GHz.
It does not send sound and it does not depend on visible light. The radar transmits radio waves, receives reflections, and analyzes the returned signal.
Why Wavelength Matters
Wavelength is:
wavelength = c / frequency
Typical values:
24 GHz: about12.5 mm60 GHz: about5 mm77 GHz: about3.9 mm
Short wavelength helps with compact directional antennas and makes small displacement visible as phase change. That is one reason mmWave can observe breathing and other tiny motions.
FMCW Measures Range
Many presence radars use FMCW, frequency-modulated continuous wave. The radar transmits a chirp whose frequency changes over time. The echo returns slightly later. Because the transmit frequency is changing, the delayed echo creates a beat frequency.
The first-order relationship is:
R = c * f_b / (2 * S)
Where:
Ris rangecis speed of lightf_bis beat frequencySis chirp slope
The factor 2 appears because the signal travels out and back. This step separates echoes into range bins.
Doppler Measures Velocity
If a target moves toward or away from the radar, the echo phase changes across chirps. The radar can estimate radial velocity.
A common relationship is:
v = f_d * wavelength / 2
Where:
vis radial velocityf_dis Doppler frequencywavelengthis radar wavelength
Radar is directly sensitive to radial velocity. Lateral movement can be much weaker if it has little component toward or away from the radar.
A Still Target Does Not Have Strong Doppler
Moving targets are easy to understand: if a person walks toward or away from the radar, Doppler and phase changes indicate velocity.
But a perfectly still target has near-zero radial velocity. It does not produce strong ordinary Doppler. A wall, a desk, and a completely still person can all sit near zero velocity.
Static presence detection works because a human body is rarely perfectly static. Even when sitting still, people have:
- Breathing motion
- Small posture adjustments
- Muscle micro-movements
- Heartbeat-related tiny vibration
- Clothing and body-surface changes
The radar may observe those as amplitude and phase changes in a range/angle region.
Why a Human Is Not Just a Static Object
The difference between a person and a wall is not that only a person reflects mmWave. Walls, desks, chairs, and metal frames all reflect.
The difference is that a human body is hard to keep perfectly stable. Breathing, posture maintenance, head or hand movement, clothing, and slow body changes create a reflection pattern that changes over time. The radar sees a group of reflections from body parts, clothing, furniture, and nearby structures.
The algorithm decides whether that region looks more like human presence than stable background.
Phase Makes Tiny Motion Visible
A small target displacement changes the round-trip path length and therefore the echo phase. A simplified relationship is:
phase_change = 4π * displacement / wavelength
At 60 GHz, wavelength is about 5 mm. Millimeter-level breathing motion can produce measurable phase variation.
This does not make phase a magic answer. Phase is affected by noise, multipath, antenna coherence, target posture, range-bin selection, and filtering. It is a useful signal, not direct semantic understanding.
Background Modeling Is Essential
An indoor radar always sees reflections from walls, doors, desks, cabinets, glass, and metal frames. If it only asked “is there reflection here?”, the answer would always be yes.
Presence detection has to ask:
Which echoes are stable background?
Which echoes are new or persistently changing targets?
That is why multi-frame background modeling is central. It also creates engineering limits:
- If someone is present during learning, the background can be polluted
- A person who stays extremely still may gradually lose confidence
- Furniture movement changes the learned scene
- Fans, curtains, air vents, and plants may look like persistent motion
If a person is extremely still for long enough, some algorithms may gradually reduce confidence or absorb that region into the background. That is a boundary of the detection definition, not necessarily a hardware fault.
Angle Estimation Is Coarse
Many mmWave modules estimate angle using multiple transmit and receive antennas. The same echo reaches antennas with slightly different path lengths and phases. Those phase differences provide direction.
Phase difference across antennas -> target direction
This helps split a room into zones, but it is not camera-like geometry. Resolution depends on antenna count, aperture, SNR, and algorithm design.
False Alarms and PIR Fusion
mmWave is robust to light and can work behind some plastic covers, but it does not understand a room. False alarms often come from:
- Fans and oscillating air vents
- Moving curtains and plants
- Pets and robot vacuums
- Multipath from metal or corners
- Radar module vibration
- Poor background-learning timing
Many battery-powered presence sensors combine PIR with mmWave. PIR provides low-power evidence of human motion; mmWave maintains presence using range, phase, and micro-motion.
PIR: strong evidence for entry or motion
mmWave: ongoing evidence for still presence
Background and zone logic: reduce fan, curtain, and plant false alarms
PIR can suppress some non-human motion false positives, but it cannot replace mmWave for still presence because a still person may stop triggering PIR.
Missed Detections
Missed detections happen when the human feature becomes too weak or does not match the algorithm’s assumptions.
Common cases include:
- Breathing motion not facing the radar well
- Body blocked by a desk, chair back, bed frame, or blanket
- Target too far away
- Person near the edge of the antenna field of view
- Algorithm thresholds tuned conservatively to reduce false alarms
- Long still periods causing confidence to decay
“Supports static presence” does not mean every posture, range, and environment will work.
Difference From Ultrasonic Ranging
Ultrasonic ranging and mmWave radar can both report distance, but their first-layer measurements are different.
Ultrasonic sensors measure acoustic flight time and depend on sound speed, acoustic reflection, and target surface.
mmWave radar measures electromagnetic echoes and can also use Doppler, phase, angle, and micro-motion. That makes it suitable for presence detection, but also introduces radar-specific multipath and background-modeling problems.
They are not interchangeable just because both can report range.
Engineering Takeaway
mmWave static presence detection is useful, but the word “static” should not be overread.
The radar does not see a human label. It sees echoes and changes. Human presence is inferred from range, angle, energy, phase, micro-motion, and background history.
Short wavelength makes tiny motion observable.
FMCW separates echoes by distance.
Phase and multi-frame changes expose micro-motion.
Algorithms interpret those patterns as presence.
That interpretation can fail when the scene produces similar echo changes or when the human signal becomes too weak.