Font Size: a A A

Research On Mobile Phone Identity Authentication Technology Based On Fuzzy ART Algorithm

Posted on:2021-03-23Degree:MasterType:Thesis
Country:ChinaCandidate:Y F ZhaoFull Text:PDF
GTID:2518306197997119Subject:Electronics and Communications Engineering
Abstract/Summary:PDF Full Text Request
Smart phones make people's lifestyles convenient and fast,but the ensuing problems such as virus intrusion,dictionary attacks,information leakage,etc,make mobile phone identity authentication based on accounts and passwords no longer safe and reliable.In response to the above phenomena,various alternative solutions have emerged,such as fingerprint identification,face recognition,iris recognition and other biometric authentication methods.Although the accuracy of these authentication methods is high,the fault tolerance is poor,and the special hardware increases the cost of the mobile phone.Therefore,there is an urgent need for a safe,easy-to-use and low-cost mobile phone identity authentication method.In order to solve this problem,based on the user's keystroke characteristics,this paper proposes a mobile phone identity authentication scheme based on Fuzzy ART algorithm to enhance the security of account and password authentication.Through the study of keystroke identity authentication,a total of three problems were found: first,when extracting feature values,because the type of the selected feature value is single,the classification model does not fit the keystroke feature well;second,In the data processing,due to the large number of keystroke texts and the lack of data optimization,it leads to the problem of too long learning and training time;third,during pattern training,due to the impact of the number of training samples and training times on pattern learning,The problem of inaccurate test identification.In view of the above phenomenon,this paper designs an identity authentication method based on Fuzzy ART algorithm.First,select the keystroke time and acceleration as the combined features to increase the type of keystroke features and avoid the lack of diversity learning of the keystroke features by the classification model;second,vector design the data filtered by the Z-score standardization method,After reconstruction,input to the classification model for regular learning,so as to improve the timeliness of training;Finally,the experiment is used to appropriately adjust the warning threshold of the Fuzzy ART classification model,and control the number of samples and the number of trainings of its model training,so as to Enhance the stability of the classification model to make up for the shortcomings of previous model training experiments.The above design scheme is simulated by MATLAB,and compared with the four commonly usedclassification algorithms,and the three performance evaluation indicators are used to analyze.The authentication method designed in this paper has a false acceptance rate of 5.67% and false rejection rate for the test sample It is 6.35% and the equal error rate is 5.87%,which is stronger than the traditional Euclidean distance,Manhattan distance,support vector machine and KNN algorithm for the classification of test samples.In summary,the research on this keystroke identity authentication scheme can effectively strengthen the security of mobile phone account and password authentication.
Keywords/Search Tags:Keystroke identity authentication, Combined features, Vector reconstruction, Fuzzy ART algorithm, Pattern training
PDF Full Text Request
Related items