Security Detection Technology For Mobile Internet Packets Based On Mining Strategies

With the fast development of mobile devices today,various malwares toward them are emerging.Due to the large number of users and rich personal information in them,it has become an important issue to protect the privacies of the users of these devices.Traditional mobile devices'anti-virus software have many defects like computing resources consumption,update delay and unable to quarantine.Aiming to solve these problems,we develop a service-oriented mobile malware detection system with the following advantages.1.It protects the mobile devices from the internet side by detecting the internet packets using data mining methods,which avoids the defects of traditional methods.2.The system is service-oriented,which is deployable by the mobile operators,and sent alarms to the users who have malware on their devices.3.The system is based on Hadoop to perform parallel computing to respond in time with the large volume of mobile internet traffic.4.Several detection modules of different algorithms including Naive Bayes,Decision Tree are implemented to ensure the quality of the detection results.In addition,to further improve the efficiency of mining,we proposed a novel algorithm,namely contraction clustering,which has two versions for different situations.It has the following advantages.1.It can effectively reduce the size of the input data,thus reduce the complexity and improve the accuracy as well.2.It is implemented using MapReduce programming model to be consistent with our system for better performance and scalability.