Gait-Based User Classification Using Phone Sensors
The authors investigate whether smartphones can be used to distinguish different users based on their gait, the rhythmical body movements of human beings as they walk. To this end, they propose, describe, and experimentally evaluate a system that classifies peoples' gait patterns using the tri-axial accelerometer of the Motorola Droid phone. The system employs the wavelet transform to extract features from raw acceleration data and the k Nearest Neighbors (kNN) algorithm to perform the classification. Preliminary experimental results show that the system achieves high classification rates (i.e. above 90%) when users walk at approximately constant speeds regardless of variations in environment.