Wireless Sensing¶

Wireless sensing starts from a simple fact: people and objects change wireless signals. When a person stands up, turns, walks, or moves away from a device, the propagation paths, reflections, phases, and amplitudes change.

mmLock uses mmWave FMCW radar, but it belongs to a broader family that also includes WiFi CSI sensing, UWB, RFID, acoustic sensing, and radar sensing.

Shared Structure¶

Most wireless sensing systems follow this path:

transmit a known signal
-> the signal interacts with people and the environment
-> receive the changed signal
-> extract features
-> infer position, motion, or state

The measured changes may include amplitude, phase, Doppler-related frequency shifts, or multipath structure.

Why WiFi Sensing Feels Similar¶

WiFi CSI describes the wireless channel between transmitter and receiver. A moving body blocks, reflects, or scatters WiFi signals, so CSI amplitude and phase change over time.

FMCW radar also transmits a signal and reads the changed return. A human body reflects mmWave signals, and the return contains range, velocity, angle, and power information.

Aspect	WiFi CSI	FMCW Radar
Uses wireless propagation	Yes	Yes
Affected by body movement	Yes	Yes
Affected by multipath	Yes	Yes
Can detect presence or motion	Yes	Yes
Needs signal processing and modeling	Yes	Yes

At the high level, both systems observe that people move and wireless signals change.

The Important Difference¶

WiFi is primarily a communication system. Sensing with WiFi means extracting behavior clues from communication-channel measurements.

FMCW radar is designed for sensing. Its chirps make range, velocity, and angle estimation much more direct:

raw ADC
-> Range FFT
-> Doppler FFT
-> Angle FFT
-> point cloud
-> sequence model

This is why mmLock is built around high-quality mmWave radar imaging rather than generic wireless disturbance detection. The task is not only to know that something moved, but to understand whether the user is leaving the device.

Limits¶

Wireless sensing is not magic. Multipath, multiple people, body pose, sensor placement, and environment changes all matter. A good document should keep mmLock tied to its actual task: user-leaving detection for data protection.