The Risk Of User Privacy Security From Insensitive Sensors On Smartphone Terminal

With the widespread use of mobile internet, mobile terminal represented by smartphones is more and more widely used in our life. As the rapid development of sensing technology and various embedded motion sensors, today’s smartphone not only serves as the key computing and communication mobile device of choice, but it also comes with a more intelligent, user friendly features. Mobile sensors can detect the changes of distance, temperature, light, pressure and other parameters. According to permissions required when installed in smartphones, these sensors can be divided into two parts, sensitive and insensitive sensors. Sensors such as GPS, microphone, camera and Bluetooth are sensitive sensors since accessing these sensors requires security permissions. Sensors such as accelerometer, magnetometer and gyroscope are insensitive since accessing them don’t need any permission.Today’s smartphones are equipped with various embedded motion sensors which has potential risks of leaking user’s private information. Recent research on smartphone sensors focus on Activity Recognition, User Input Inference and Location privacy, paid less attention on privacy leaking. To solve this problem, this article describe in detail the key contributions based on smartphone sensors:(1) Firstly we analysis security mechanisms of the current smartphone platform and sensors, then we present the design and implementation of SensorLogger, an Android application that used to collect sensor readings, and the power consumption problem that user concerned.(2) Based on the way user holding and touching the screen of smartphone, we study the feasibility of predicting end user’s private information by collecting sensory data such as accelerometer, magnetometer, orientation and gyroscope from different demographic. Then we use some repair algorithm to renovate missing data and data noise. Using Weka which integrated multiple classification algorithms to create user privacy identify model. User studies show that insensitive sensors can achieve high accuracy in identifying user private information such as gender, age and occupation, then we analysis the safety performance of every sensor based on their accuracy.(3) Firstly we introduce a new mobile system framework which use passive sensory data to identify users based on biometric authentication. The framework contains sensory data collecting, data preprocessing, specific features extracted and pattern matching identify process. To collect user motion data, we modify the count and sample frequency of SensorLogger which simultaneously sampled accelerometer, gyroscope and magnetometer based on the trajectory of user "Mobile Device Picking-up" gesture. To investigate the effect of extra body movement and gesture instar on the stability of the biometric modality, we separate our data into three sessions. Since the resource and sample limit of mobile terminal, we evaluate the performance of our system using Support Vector Machine and finally carried out the whole process of user identify authentication.