Research On Mobile User Behavior Pattern Mining And Online Identification Strategy Based On User Trust

Mobile cloud computing is a new application model with the continuous development and integration of cloud computing and mobile internet.In the architecture of trusted mobile cloud services,the determination and identification of user's identity and behavior will be an important prerequisite for subsequent services.On the premise of ensuring the user's identity and user's reliability and security,according to the demand for personalized service and resource consumption of the user,the intelligent decision making service mode and the fast innuendo or dynamic allocation of service resources.Obviously,the user behavior discrimination is very important to the mobile cloud service.In this thesis,the user trust refers to the user's operations of the service implementation and actions are always within the scope permitted by the user's rules,and the system can identify and control the abnormal behavior of users.With the help of the tool of pattern mining,we can find out the behavior patterns of a large number of behavior data from the legitimate mobile users when they are using mobile cloud services,and then the reference set of normal behavior is constructed.After that,the real-time system compares the user behavior and normal behavior,to detect whether the user is abnormal,aim to achieve the purpose of supervision and recognition of user behavior,and protect the user layer data read-write environment and safety and service delivery.This thesis focuses on the user trust issues,behavior pattern mining and online identification.1.The user defined and specialized timing behavior sequence is basis of abnormal user behavior identification.First,we give a formal description of the sequence of the user's temporal behavior,and the coding structure of the sequence of user behavior.Then,the coding structure of the user's sequential behavior sequence is given,and the meanings and uses of each part in the coding sequence are introduced.2.An adaptive user normal behavior pattern coding method is proposed.In the process of user's normal behavior patterns,According to the characteristics of mining temporal behavior patterns,we draw on the theory of genetic algorithm,and use an adaptive coding method to build a complete set of normal user behavior patterns to improve the efficiency of real time judgment of the user's sequential behavior.Simulation results show the advantages of the proposed algorithm in terms of both dynamic performance and operational efficiency.3.Based on the collection of the completely normal behavior pattern of the user,we design a real-time identification method of the user's temporal behavior based on the Needleman-Wunsch algorithm.The algorithm transforms the real-time behavior of users into formal coding,and then compare it with the normal behavior pattern set,determine whether the user behavior is abnormal by the threshold set in advance.Then some details of anomaly detection are discussed.Based on the theoretical research results,we carry out the simulation,and the results show that the detection algorithm presented in this thesis is superior to the traditional algorithm in detection rate and false positive rate.The innovations of the pattern-mining algorithm and online anomaly detection algorithm of this thesis are listed below.After encoding the behavior sequence,it has a high similarity with DNA sequence,so this ` thesis innovatively introduces the genetic algorithm and DNA algorithm in biology sequence alignment of pattern mining and anomaly detection.At the same time,due to the complexity of mobile cloud user terminals and the changeable access environment,mobile cloud computing servers are faced with more diverse and more frequent abnormal attacks.If the same pattern-mining algorithm and parameter model are used,the accuracy and false positive rate of anomaly detection will fluctuate greatly.For this reason,this thesis designs an adaptive encoding method and parameter determination algorithm,which can make the system adjust its own mining and detection strategies according to the change of the environment,and improve the operation efficiency and detection performance.