| Summary: | We collected inertial sensor data on a smart ring and a smartwatch, both worn on the same arm, and from a Raspberry Pi IMU mounted on a closed door as users performed activities. The data contains data from users (n=21) as they performed five types of gesture: tap gestures as users performed mobile payments on a number of point-of-sale terminals with (i) the smart ring and (ii) the smartwatch, and knock gestures as users knocked on the door in patterns of (iii) 3 knocks at a time, (iv) 5 knocks at a time, and (v) secret knocks chosen by the user. All of these gestures can be used to authenticate the user.
Each user folder contains 1 nfctimstamper.csv file, 1+ <date>-<time>-watch-sensors.csv file(s), 1+ set(s) of <date>-<time>-ring-*.csv files, and 1+ <date>-<time>-door-sensors.csv file(s). These types of files are described as follows:
- nfctimstamper.csv
This file contains the timestamps collected for the purpose of segmenting the watch, ring, and door sensor data. The first column is the human-readable timestamp, the second is the UNIX timestamp, and the third contains the data read from the NFC tag that triggered the entry. The NFC tag is either 'userWTC' which was attached to the smartwatch, 'userRNG' which was attached to the smart ring, or something containing 'TAP' which shows which point-of-sale terminal is being interacted with -- that is, the 'TAP' tag tells the system which terminal is being used and then each subsequent 'userWTC' or 'userRNG' is the timestamp of the NFC connection established during a tap gesture as the user interacts with that terminal. The third type of tag shows when a session of knocking gestures has been initiated, either 'NOC3', 'NOC5', or 'NSEC', representing the 3-knock, 5-knock, or secret knock gestures, respectively. To identify the type of knock gesture that is being performed in between button-presses given in a <date>-<time>-ring-buttons.csv file, cross-reference the timestamp of the starting button-press with the preceeding knock gesture tag. The fourth type of tag is of the form 'ATK-<gesture>-<victim>', which timestamps an impersonation attempt made by the user having watched a video clip of <victim> performing <gesture>.
- <date>-<time>-watch-sensors.csv
These files contain the inertial sensor data collected on the smartwatch. Use the nfctimestamper.csv file to segment this data into tap and knock gestures.
- <date>-<time>-ring-sensors.csv
These files contain the inertial sensor data collected on the smart ring. Use the nfctimestamper.csv file to segment this data into tap and knock gestures. The timestamps in the 'timestamp_us' column are local to the smart ring and given in milliseconds; use the associated <date>-<time>-ring-timesync.csv file to synchronise these timestamps.
- <date>-<time>-ring-buttons.csv
This file contains the timestamps of the button-presses at the start and end of each knock gesture for the purpose of segmenting those gestures. Each button-press action caused multiple events to be logged in quick succession, typically three at a time; for consistency, only the first timestamp should be used and the others discarded.
- <date>-<time>-ring-timesync.csv
This file contains the local timestamp of the smart ring at the time of the creation of the file and the UNIX timestamp taken at the same time for the purpose of synchronising the timestamps in the associated <date>-<time>-ring-sensors.csv file with all other files.
- <date>-<time>-watch-sensors.csv
These files contain the inertial sensor data collected by the Raspberry Pi attached to the door. Use the nfctimestamper.csv file to segment this data into knock gestures.
|
|---|