User Implicit Authentication And Application Recognition For Intelligent Devices Based On Sensor Data

With the continuous development of mobile Internet and intelligent devices,mobile intelligent devices are used in all aspects of life,and become an indispensable part in daily life,entertainment,work,learning and mobile payment.A large number of users' privacy data are stored in intelligent devices,and the authentication system has become an important defense to protect users' privacy data.However,a large number of side channel attacks threaten the traditional explicit authentication,such as side channel attacks based on computer vision and touch screen attacks.The user's authentication key is facing the risk of leakage.In addition,the development of mobile applications makes the perception of users' running applications become a demand: advertising operators need to perceive users' application usage to make decisions on advertising resources;In some restricted areas such as confidential units,examination rooms,it is necessary to monitor the running applications of users' intelligent devices to restrict their use of some applications such as photos,browsers.In order to solve the above problems,this thesis proposes the technology of user implicit authentication and application recognition based on sensor data.In the login phase,users are pre authenticated based on implicit authentication to prevent related side channel attacks and improve the security and convenience of authentication;In the user's session phase,the application that the user is running is identified.The specific work of this thesis includes the following three aspectsFirstly,this thesis proposes the user implicit authentication technology based on physiological behavior characteristics,which uses the most commonly used action in daily life to pre authenticate the user in the process of taking the smart device.The existing sensors of mobile intelligent devices are used for sensor data fusion to obtain the time series data that can reflect the user's finger size,arm movement speed,and height physiological behavior characteristics.Finally,the user is implicitly authenticated through the authentication model.Secondly,this thesis proposes the application recognition technology based on the channel analysis of the magnetic field data side,which uses the magnetic field sensor to collect the magnetic field data around the user when using the intelligent device,and realizes the application recognition for the intelligent device by analyzing the magnetic field data.In the process of human-computer interaction,the magnetic field data is collected and preprocessed to get the magnetic field data when the intelligent device application starts.The characteristics of the magnetic field data are extracted,and the machine learning algorithm is used to classify,and then the running applications of the intelligent device are identified.Finally,this thesis designs user implicit authentication and application identification system.Integrating the above theoretical and technical achievements,on the premise of ensuring availability and security,we design and develop a prototype system of smart device user implicit authentication and application identification based on sensor data,and test the function and performance of the prototype system to verify the feasibility and effectiveness of the system.To sum up,this thesis proposes user implicit authentication technology based on physiological behavior characteristics and application recognition technology based on magnetic data channel analysis.A prototype system of user implicit authentication and application recognition for intelligent devices based on sensor data are designed.The research work can provide technical support for the authentication security and application identification of intelligent devices.