Abstract
At an ever increasing rate, the smartphones and other devices people carry with them in their everyday lives are packed with sensors and processing power. This provides an unprecedented opportunity to apply data mining techniques to people’s activities as they go about their daily lives, without changing their routine. The goal of the Wireless Sensor Data Mining (WISDM) Project is to explore the possibilities of data mining on these powerful mobile platforms.1 Data mining involves extracting knowledge from data using computer algorithms. A major sensor in these devices is the tri-axial accelerometer originally included for screen rotation and advanced gaming. Our work, so far, has focused on using data mining methods on the accelerometer data to identify the activities users are performing (activity recognition) while carrying the phone. Many useful applications can be built if accelerometers can be used to recognize a person’s activity. We have also demonstrated that accelerometer data can be used to uniquely identify and authenticate users. While some previous work has examined sensor-based gait recognition,2-12 our work in this communication differs in that we identify users based on the way they move during multiple activities (i.e., not just walking) using only commercially available smartphones, which are carried in the user’s pocket.
Recommended Citation
Lockhart, Jeff FCRH '13
(2013)
"Mobile Sensor Data Mining,"
The Fordham Undergraduate Research Journal: Vol. 1:
Iss.
1, Article 11.
Available at:
https://research.library.fordham.edu/furj/vol1/iss1/11