Research And Implementation Of Continuous Authentication For Smart Device Users

Smart devices are widely used in our daily life,work and study environment nowadays,involving shopping,catering,accommodation,transportation,etc.And they are becoming the attack targets of various attackers due to the users' considerable private and confidential data stored in these devices.As authentication is the primary defense technique for keeping users'information secure,various side channel attacks,such as touchscreen residue based attacks,computer vision based attacks and built-in sensor based attacks,are proposed to compromise the existing authentication techniques so as to obtain the authentication credentials.Furthermore,most authentication techniques adopted by mainstream smart device platforms only authenticate the user's authenticity at login phase.This makes the system vulnerable against attacks during authorized sessions.Therefore,the users' information lacks full protection.To address these issues,a continuous user authentication scheme is proposed for smart devices during both login phase and post-login phase.On one hand,we study and design a Back of Device(BoD)tap based secure explicit authentication technique for user login phase to defend against side channel attacks.On the other hand,we study and design a 3D magnetic finger motion pattern based secure implicit authentication technique to remedy the drawbacks of existing authentication techniques.Specifically,our work includes the following aspects:(1)We study and design a BoD tap based secure explicit authentication technique.For cursor movement control,we design a novel tilt,based secure cursor movement control algorithm to increase the trajectory space of the cursor.For BoD tap dectection,by leveraging the built-in accelerometer,we study and design a machine learning classification algorithm based BoD tap detection technique.Finally,we achieve the secure explicit authentication through combining the cursor movement control algorithm and the BoD tap detection technique.(2)We study and design a 3D magnetic finger motion pattern based secure implicit authentication technique.By introducing a magnetic ring,we propose a built-in magnetometer based three dimensional finger motion pattern analysis scheme without any hardware modification of the smart device.During user-device interactions,we implicitly record the built-in magnetometer data.After the cancellation of background magnetic field,we analyse the value changing pattern of the magnetic field which caused by the motion of a ring-wearing finger around the smart device.Then,we apply machine learning classification algorithm to achieve the implicit user authentication.(3)By integrating the theoretical and technical research findings,we implement the BoD tap based secure explicit authentication technique and the 3D magnetic finger motion pattern based secure implicit authentication technique.We further develop a continuous user authentication prototype system for smart devices by combining the two techniques,and we also conduct functional and performance tests,which verified the feasibility and effectiveness of the system.In summary,to mitigate various side channel attacks,this thesis investigated and designed a BoD tap based secure explicit authentication technique.To alleviate the drawbacks of the existing authentication techniques,this thesis studied and designed a 3D magnetic finger motion pattern based secure implicit authentication technique.Finally,a continuous user authentication prototype system for smart devices is designed and implemented to provide full protection for users' information.