Enhanced mobile malware detection using intersection attributes technique

• Main Author: Ismail, Najiahtul Syafiqah
• Format: Thesis
• Language: English; English
• Published: 2022
• Subjects: T Technology (General); TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://eprints.utem.edu.my/id/eprint/26919/1/Enhanced%20mobile%20malware%20detection%20using%20intersection%20attributes%20technique.pdf; http://eprints.utem.edu.my/id/eprint/26919/2/Enhanced%20mobile%20malware%20detection%20using%20intersection%20attributes%20technique.pdf; http://eprints.utem.edu.my/id/eprint/26919/; https://plh.utem.edu.my/cgi-bin/koha/opac-detail.pl?biblionumber=122085

The user-friendly interface, easy to use, and have many alternatives for the applications market make Android one of the most popular smartphone operating systems in this 21st century. In a mobile application, permission is one of the essential elements to protect user’s personal information and privacy. Permission-based detection has been used widely but is deemed insufficient because it still suffers from high false alarm rates due to the permission-based issue. The current detection technique generates high false alarm rates, making the detection technique less effective in detecting the permission-based attack. Therefore, this research aimed to improve permission-based detection by integrating permission attributes with intent. However, integrating multiple attributes will increase the number of attributes used in mobile malware detection and affect the false alarm rate. Thus, the optimal size of attributes was developed to reduce the high false alarm rates generated by the Mobile Malware Detection System. Hence, this research introduces an Intersection Attribute Technique to reduce the number of attributes generated and improve the quality of attributes selected in the attribute selection process. The proposed technique with the Venn Diagram concept determined the correlation between attributes in the same process during Pre-processing phase before undergoing the Correlation Feature Selection process. Support Vector Machine was used to classify the applications. A comparative analysis has been performed using the proposed approach and three other approaches. The dataset used in this research is from New Brunswick Repository. The result indicates the Intersection Attribute Technique can reduce the number of attributes generated by 18, accuracy 96.67% and the false positive rate is 0.04%. In conclusion, the Enhanced Mobile Malware Detection with Intersection Attribute Technique can classify benign and malicious mobile applications more accurately and minimize false alarms.