Table 1.
Features that can be extracted from the smartphone’s sensors to characterise physical movement. The smartphone’s camera is not listed in the table as the images they can produce are often processed with “computer vision” based methods, e.g., speeded-up robust features [66].
| Domain | Feature | Accelerometer | Gyroscope | Magnetometer | Barometer | Microphone | GPS/GSM/Wi-Fi |
|---|---|---|---|---|---|---|---|
| Position/altitude | ✓ | ✓ | ✓ | ✓ | ✓ | ||
| Time | Signal Magnitude Area [2] | ✓ | ✓ | ||||
| Signal vector magnitude [4] | ✓ | ✓ | ✓ | ||||
| Differential pressure [11] | ✓ | ||||||
| Autoregressive coefficients [56] | ✓ | ||||||
| Tilt angle [58] | ✓ | ✓ | |||||
| Relative altitude [67] | ✓ | ||||||
| Peak-to-peak amplitude | ✓ | ✓ | |||||
| Zero crossing rate [68] | ✓ | ✓ | ✓ | ||||
| Short-time average energy [69] | ✓ | ||||||
| Frequency | Low energy frame rate [46] | ✓ | |||||
| Entropy [64] | ✓ | ✓ | |||||
| Energy [70] | ✓ | ✓ | |||||
| Fast Fourier transform coefficients [71] | ✓ | ✓ | |||||
| Discrete cosine transform coefficients [72] | ✓ | ✓ | |||||
| Spectral flux [73] | ✓ | ||||||
| Spectral roll-off [74] | ✓ | ||||||
| Spectral centroid [74] | ✓ | ||||||
| Bandwidth [74] | ✓ | ||||||
| Normalised weighted phase deviation [75] | ✓ |