Researchers develop software to curb kids' obsession with smartphones

If you give your smartphone to an elfish kid, you may encounter the following scenarios. A two-year-and-a-half old boy who wants to watch an animated film tries passcode after passcode like crazy. As a result, the phone will be “locked” for 48 years. A 7-year-old British girl cracks the password of her parent’s phone and spends over ￡10,000 in booking a package tour of the Disneyland.

After numerous tussles with their impish kids, parents may hope to have a smartphone which can identify a child user in an automatic and friendly manner.

Fortunately, this idea may well be realized in the not-too-distant future.

Recently, a research team led by Prof. XU Wenyuan and Prof. JI Xiaoyu with the Ubiquitous System Security Laboratory of the College of Electrical Engineering, Zhejiang University, worked in collaboration with the University of South Carolina and developed iCare to identify childusers based on their touch behaviors in an automatic and implicit way.

Researchers extracted and exploited two key features, i.e., hand geometry and finger dexterity that are represented in 35 subfeatures, to capture the distinct characteristics of children. A dataset on 31 people including 17 children (3 to 11 years old) and 14 adults (22 to 60 years old) was collected and tested to verify the effectiveness of iCare through two games—entering the PIN number and playing a popular puzzle game called “2048”.

“These two games involve tapping and swiping, which are readily accessible for children,” said JI Xiaoyu.

Data related to 35 features, including the frequency, size, velocity, trajectory distance and direction of tapping and swiping, were extracted.

Although only a few subjects were aged between 3 and 5, their frequency of tapping was the highest among 4 groups. This may be attributed to the fact that the probability of their entering a wrong password was the highest, which constituted one of the key determinants in child identification.

Existing methods for parent patrol are far from satisfactory. “We attempt to utilize built-in sensors which don’t impose privacy violations so as to achieve spontaneous identification,” JI Xiaoyu explained, “Data derived from tapping and swiping behavior is differentiated from privacy-related vocal and visual data. This way, we can identify user features in a precise and automatic way without the invasion of privacy.”

Results showed that iCare can achieve 84% accuracy for child identification using only a single swipe on the screen, and the accuracy rate can hit 97% with 8 consecutive swipes.

Much to parents’ relief, kids are incapable of breaking through the “defense” against natural recognition because their finger size is far smaller than that of an adult and they are unable to develop a similar swiping habit as adults. Even if they may happen to succeed once, it is still a “mission impossible” to satisfy 35 features for an unfledged kid.

“Upon detection of a child user, the system will give an automatic reminder of the remaining usage time. When time is up, the phone will be automatically switched off. We can also design two systems and develop secure and child-friendly apps or websites,” said JI Xiaoyu.

“The automatic identification technology such as iCare shows great promise. We look forward to working together with public services agencies, like internet-based enterprises and hospitals, and expanding channels for corresponding service technology for specific groups,” said XU Wenyuan.