Paper Details

The attraction of users towards Android mobile devices is increasing day by day due to its ubiquitous characteristic like easy accessing of information from anywhere and anytime. The availability of open app market systems, leads to the cause for the escalation of malicious Android apps. The increasing usage of mobile device leads to development of new applications, which ends up with more number of application creation for all day to day transactions, other activities and entertainment purposes. The complexity and variety of the vindictive Android applications render the traditional malware recognition strategies ineffectual, which brings about an enormous number of malevolent applications staying undetected. For this reason it is necessary to have an irresistible technique for identification and classification of Android malicious applications. Hence, in this proposed work, a mechanism is presented, called malign android application detection system using Decision tree algorithm and Support Vector Machine. Since the launch of the smartphones and it has become an important part of our lives. Now a day’s peoples are very much dependent on smartphones applications for their entertainment reasons, the huge several of applications are downloaded by the users from play store app or from the trusted third party applications. Also due to the unawareness of users, the applications may be downloaded from unauthorized sources pose a threat as it doesn’t undergoes the necessary checks or mechanisms to validate the authenticity of these applications and maybe infected with malware. The malware injected applications creates a hole to the flow of user’s personal information or to obtain unauthorized access to the system. In this work, we proposed an auxiliary mechanism using C-S DTFS algorithm for malicious app identification by using machine learning techniques for classification of the applications as either malicious or benign and the obtained results are compared to identify the best suited algorithm for our dataset. The issues in existing work are that they used Random forest and Naive Bayes Algorithm. The main issue of random forest is that if number of trees are more in number, will lead to the too slow execution of algorithms results in unsuccessful predictions.

C-S DTFS, Information Gain, Support Vector Machine, Decision Tree