The User Abnormal Behavior Detection Under Mobile Cloud Service Environment

Mobile cloud computing integrates cloud computing and mobile Internet,which is a new application mode.Mobile cloud computing overcomes the limitations of mobile terminal hardware,more readily satisfied in portable data accessing and intelligent load balancing.However,it's more likely to lead to security problems in its “user-environment-service” levels.Therefore,how to identify the identity of the users and their credibility before the cloud services enter into the substantial service process has become the core issue.From the perspective of user credibility,we research the user abnormal behavior analysis method in mobile cloud environment progressively to solve above problems.The major contributions of this thesis can be summarized as follows.1.In the process of traditional clustering analysis and calculating similarity,there always are the problems of over-fitting and flooding the feature information,as a result,we propose a user abnormal behavior analysis method based on neural network clustering.In this method,we use singular value decomposition(SVD)to reduce dimension and de-noise for massive data.Besides,we add information entropy to hidden layer of neural network model for softening points.Moreover,we get the weight of each attribute by information entropy to calculate the similarity.The simulation results show that the scheme has higher detection speed and clustering accuracy than traditional schemes.The proposed method is more suitable for the mobile cloud environment.2.In order to solve the unbalanced data learning problem caused by the unequal sampling rate in research content 1,we propose a collaborative analysis method of user abnormal behavior based on reputation voting.This method combines anomaly detection and misuse detection technology,and uses under-sampling and pruning technique to construct training samples.Then we use ensemble classifiers voting user behaviors to identify abnormal behavior which based on reputation value.The voting mechanism according by principle of minority is subordinate to the majority,which improves clustering accuracy and detection speed.The experimental results show that the scheme has better effect for larger dataset with unbalanced rate especially.3.In order to identify abnormal behavior and user intention in advance,we propose an abnormal behavior recognition and autonomous optimization method based on pattern-growth.We use the method of real-time distinguishing user behaviors based on hierarchical matching,aiming at judging whether user behaviors beyond trusted tolerance range.Moreover,we analyze function and data flow to determine abnormal behavior.Finally,we use pattern growth algorithm to build the complete normal node subgraph sets and abnormal affect node subgraph sets,aiming at independently judging user behaviors in advance.To identify abnormal behavior in the mobile cloud environment has become a core issue in the field of mobile cloud computing.This thesis mines and analyzes the behavior patterns of users based on their history behavior data.Thus the abnormal behaviors can be analyzed and identified according to the existed behavior patterns.The research methods of this thesis guarantee that the implemented operations to service by users are within the allowable rule range of users,which laid a solid foundation that mobile cloud service provide customers with low consumption,high efficient and reliable service.