The Research On Pattern Password Based Behavioral Characteristics Recognition

Smartphones are often used to make purchases, payment or storing all kinds of private information. Once a phone is lost or user password is cracked, the user will have the huge losses. Therefore, the identity authentication mechanism to identify your smartphone is very important. Currently there are three commonly authentication mechanisms on smartphone:one is the PIN code; the second is the pattern password; third, and the latest one is Fingerprint recognition. Because the fingerprint identification requires specific hardware, it is not universal in the smartphone market. The pattern password is the most simple and fast user authentication mode. It is widely used in smartphones and other mobile devices, commonly used in the lock screen or lock the phone application that can be used to prevent unauthorized users from operating the mobile phone device or applications.Users tend to set unlock pattern or PIN code too simple, they are easily guessed. Further studies have shown that, you can quickly crack the password by extracting the oil traces on the phone screen. Therefore, it can't solve the password leaking relying solely on the existing password verification mechanism. So, we propose a biological behavior characteristics authentication mechanism. The common Biological behavioral recognition include gait recognition, handwritten signature recognition and keystroke behavior recognition. The basic principles of these mechanisms is that everyone's behavioral is special. You can extract the user's behavior data as everyone's fundamental characteristics of the user to verify his or her identity. Based on this principle, this paper present a behavioral recognition mechanism based pattern password on smartphones with Android system.Because smartphones with numerous sensors, you can collect user's behavioral characteristics data when they are operating mobile phone in anytime and anywhere, and you don't need any additional equipment. This is almost zero cost with respect to the fingerprint recognition, face recognition and other biometrics. And for users, it is completely transparent when a user enters the phone pattern password. Recognition system automatically extracts user behavior characteristic data.In this paper, we extract the behavioral characteristics data of the user unlocking pattern password, such as the position of the touching screen, pressure, contacting area, time and other information. They constitute a unique user touching behavior. Then we use support vector machines and Naive Bayes to trained user behavior model and comparative analysis.This paper has three parts, the first part includes a first and second chapter, namely the background and basis of relevant theories. Among them, the first chapter describes the smartphones and other mobile devices in people's lives play an increasingly important role, as well as popular after potential safety problems. And then briefly introduced biometric authentication, as well as existing security mechanism on smartphones. The second chapter introduces the concepts and theories used in this article. Firstly, the paper introduces the technology and features of common behavioral characteristics recognition, and then introduces constitute of unlock pattern that is the most commonly used on mobile devices; thirdly, introduces the basic concepts of the related machine learning algorithms include SVM and SMO.The second part of this paper is the third and fourth chapters, this part is the core of this paper. This third chapter introduces the experimental system model structure and the collection of detailed behavior characteristics data, including preparatory work, feature collection systems, and data processing later. Chapter IV followed the contents of previous chapter. The evaluation indexes of model were introduced at first, including accuracy, precision, recall, F measure, ROC curves and AUC and so on. And then the results of classification model were compared.The last part, which is the fifth chapter, includes summary and outlook. This part summarize the research work of this paper, and put forward the new development direction.