Description
Accelerometers (typically MEMS, often part of a 6- or 9-axis IMU) appear in the bibliography as the wearable-sensor baseline that CSI-based human-activity recognition systems are compared against. Body-worn IMUs give clean per-subject motion data but require the subject to carry a device — exactly the deployment constraint that Wi-Fi sensing is trying to eliminate. For the thesis, accelerometers are the canonical "device-bound" comparison ("CSI HAR matches accelerometer-based HAR within X% F1") and the source of ground-truth gait / activity labels in fused datasets.
Specs / capabilities
- 3-axis linear acceleration, ±2 / ±4 / ±8 / ±16 g full-scale ranges
- Sample rates 50 Hz-1 kHz typical; up to 4 kHz on premium parts
- 16-bit ADC; ~50-200 µg/√Hz noise density
- I²C / SPI digital output
- Common pairings: gyroscope (6-axis IMU), magnetometer (9-axis IMU)
Used by (papers)
- Wearable-HAR baselines compared against CSI-HAR
- Pedestrian Dead Reckoning (PDR) papers fused with BLE / Wi-Fi RSSI
- Gait analysis ground truth