Number of users = 95
Sitting - 2850 (95 x 30)
Standing - 2850 (95 x 30)
Walking - 2850 (95 x 30)
Each activity has two types of data hold-movement and touch. Hold movements is collected using seven 3-dimensional motion sensors (i.e., the accelerometer, the high-pass sensor, the low-pass sensor, the orientation sensor, the gravity sensor, the gyroscope, and the magnetometer) and touch data is collected using touchscreen sensor.
Sensor: 112 Features and 113th Column is label
Description: 4 statistical features, namely Mean (μ), Standard Deviation (σ), Skewness (s), and Kurtosis (k), that gives 16 statistical features
per sensor. Thus, a total of 16 features are obtained from a sensor. Since there are seven sensors in total so 112 satistical features are obtained.
Touch: 30 features and 31st Column is label
Description: Touch-typing features consist of 8 Type0 (timing difference between each key release and key press), 7 Type1 (timing difference a key press and previous key release, 7 Type2 (timing difference two successive keys release), 7 Type3 (timing difference two successive keys press), and 1 Type4 (timing difference between last and first key press). Thus, 30 touch-typing features from the 8-digit random-text entry are extracted.
Please cite our following papers to use the dataset.
@article{buriro2021risk,
title={Risk-driven behavioral biometric-based one-shot-cum-continuous user authentication scheme},
author={Buriro, Attaullah and Gupta, Sandeep and Yautsiukhin, Artsiom and Crispo, Bruno},
journal={Journal of Signal Processing Systems},
volume={93},
pages={989--1006},
year={2021},
publisher={Springer}
}
@phdthesis{gupta2020next,
title={Next-generation user authentication schemes for iot applications},
author={Gupta, Sandeep}, year={2020},
school={PhD thesis, Ph. D. dissertation, University of Trento, Italy}
}