Embedded systems · Implementation study
Wi-Fi presence sensing with Raspberry Pi and C++
How a compatible Wi-Fi receiver and a C++ processing pipeline can turn radio-channel changes into evidence of movement.
What the demo shows
The portfolio animation uses scripted activity and illustrative positions. It presents the Raspberry Pi radar project idea; it is not connected to a radio, and no range, accuracy, or person-count result is asserted here.
Proposed system flow
- 01Wi-Fi frames
- 02Channel measurements
- 03Windowed features
- 04Activity estimate
01
Measure the channel
Wi-Fi sensing examines how the radio channel changes. People can alter reflected signal paths, and movement can create Doppler shifts. Water content alone is not a detection method: reflections, obstructions, antenna placement, and other moving objects also matter. WiSee demonstrates wireless motion sensing in research, but its results do not establish the performance of this Raspberry Pi concept.
02
Capture useful samples
A CSI-based implementation needs a supported chipset and firmware combination that exposes channel state information, rather than only a signal-strength reading. Nexmon CSI documents one possible capture route on supported Raspberry Pi hardware. A C++ collector would validate the capture format, timestamp measurements, discard invalid subcarriers, and place samples in bounded buffers. The actual board, operating system, and firmware must be checked together before choosing this approach.
03
Estimate activity and expose uncertainty
A proposed processing stage would establish an empty-room baseline, filter noise, and summarize changes over sliding time windows. Thresholds or a trained classifier could then produce an activity estimate. Tests must include empty rooms, still occupants, fans, and changed furniture. Detecting movement does not automatically provide reliable static occupancy, head count, distance, or bearing; those require additional measurements and validation. The circular sweep and plotted dots are a display metaphor, not measured geometry.
Validation plan
- Compare labeled recordings from different days and rooms, keeping evaluation sessions separate from training.
- Report missed detections, false alarms, latency, and exact hardware configuration.
- Confirm behavior when Wi-Fi traffic stops, the receiver moves, or the environment changes.
More in embedded systems
Have a similar problem to solve?
Discuss your project