Research On Implicit Authentication Of Intelligent Devices In Mobile Social Networks

With the rapid developing of mobile Internet and the popularity of smart device,more and more identity authentication is introduced into mobile devices.At present,based on the user account and password authentication is still the mainstream mode of mobile social network authentication.This model,however,has many problems such as the user's mobile device are easily lost,stolen or borrowed.In order to solve these problems,a variety of auxiliary implicit authentication technology arises at the historic moment,such as sliding screen authentication,WeChat friends identification authentication and so on.More and more researchers begin to focus on the implicit behavioral mode of research,such as the user's gestures,keyboard,intensity of touch and so on.Furthermore,With the popularity of wearable devices,the implicit identity authentication on wearable devices also has become hot spot of research.This thesis first analyzes the advantages and disadvantages of the traditional identity authentication scheme,and deeply investigates the existing implicit identity authentication scheme based on social network.Secondly,according to the characteristics of existing implicit identity authentication,the implicit authentication of smart mobile devices and wearable devices is studied.The specific contents are as follows:Firstly,an implicit identity authentication scheme is proposed based on the touch behavior of mobile devices.The scheme consists of four parts: data acquisition,data processing,model training and certification.Taking the complexity of the touch behavior into account,in order to improve the accuracy of the certification,touch control behavior is divided into two parts of the sliding screen and keying for certification.The sliding screen is subdivided into upper slide,down slide,left slide and right slide,and mainly concerns the three characteristics of touch time,direction and speed.The keying part considers the two characteristics of the keying duration and interval.And then,combined with the idea of support vector machine,the sample data is trained and the test data is certified,the effectiveness of the new implicit authentication scheme is evaluated by the measured method.Secondly,an implicit identity authentication scheme is proposed based on the smart glasses.User's data collected by the Leap Motion equipment is used for machine learning,which means the identity authentication can be conducted inadvertently in the process of normal using of smart glasses.Bagging integration algorithm is adopted for machine learning.The error BackPropagationalgorithm,the support vector machine and the K-nearest neighbor algorithm is studied for setting base learner.According to the situation of this thesis,the error BackPropagation algorithm and the K-nearest neighbor algorithm are chosen as the algorithm of base learner.This thesis chooses the forward swipe gesture with selected fourteen features as the example,and extracted 613 data to form a training set.The accuracy of the program in the forward swipe gesture was 94.1176%.Finally,the effectiveness of our proposed scheme is indicated by simulation.