Smartphone User Identification using Bio-features (2016-2017)

Prof. Hongzi Zhu, Department of Computer Science, Shanghai Jiao Tong University • Oct 2016 — Feb, 2017

Investigated and developed a novel way of smartphone user identification using individual's biology features.

Through investigation and some experiments carried out on mobile devices and Cortex-M4 experiment board also, it is discovered that individuals have completely different tremor patterns, in both frequency and strength. Thus it should be possible to recognize the user based on his/her hand tremor and it is a novel feature that can hardly be imitated.

Experiments are carried out on both a Cortex-M4 experiment board and Nexus 4. An Android-based application is also implemented to perform user identification, in which it can collect user tremor data and then learn from them to build a classifier using SVM. It can identify the user basically, though it may go wrong sometimes. I suppose that is because the shape of user and the hardware of smartphone matters. Also, the data collected is limited.