Masquerade Detection Based on Smartphone
|School||Beijing Jiaotong University|
|Course||Computer Science and Technology|
|Keywords||Masquerade Detection Data Mining HMM SVM|
In recent years, the masquerade attack in the proportion of computer security incident is constantly increasing, becoming one of the most serious damage to the system attacking. Meanwhile with the popularity of smartphone, more users start to deposit important imformation into it, when the phone is being stolen, users more concern about the safety of the information in mobile phone. So finding out the illegal users and preventing effectively have become the issue that need to be resolved.The ways to find out the illegal user are various, and many scholars judge masquerade detection behavior on the changes of multiple performance indexes of mobile phone, but the researches on finding out illegal user by mining the user’s keyboard behavior are few. So refer to predecessors’ experience, this paper puts forward the masquerade detection based on the keyboard smartphones. In fact, the history of masquerade detection research has more than 20 years, and experiments prove that many of the algorithms in theory of statistical analysis, data mining and machine learning achieve very good results. But whether these algorithms are also suitable in masquerade detection of smartphone keyboard, whether these algorithms have very good application, and whether we have to improve it, these are the starting point of this subject. The Apriori algorithm based on data mining, the former backward algorithm and B-M algorithm based on hidden markov models, and rapid markov nuclear function algorithm based on suffix tree have been good applicable on the research of intrusion detection based on sequence. So the work of this paper is the study of the applications of these algorithms on smartphone.After the proposal, we have done the following work:1. Research of the three types of algorithm and the appropriate improvement according to the subject.2. Select training set and test set, and do the experiments of the algorithms.3. We conclude that the three types of algorithms are also applicable on the mobile terminal and find out the optimal algorithm.4. Compare with PC keyboard, pove that the three types of algorithms are more applicable on masquerade detection based on smartphone keyboard.