Журнал основ возобновляемой энергии и приложений
Открытый доступ

Абстрактный

Machine Learning for Identifying iOS Malware

Lisa Angelina

Smartphones have transformed into an indispensible component of our daily life. Smartphones are almost completely relied on as a communication tool, a source of information, and a source of pleasure on a social, political, and economic level. Rapid advances in information and cyber security have mandated particular attention to the privacy and security of smartphone data. Spyware detection systems have recently been created as a potential and appealing option for the privacy protection of smartphone users. Because the Android operating system is the most commonly used in the world, it is a major target for various groups interested in attacking smartphone users' privacy. This research presents a unique dataset gathered in a realistic setting using a novel data collecting approach based on a unified activity list.

The data is separated into three categories; Regular smartphone traffic, traffic data for the spyware installation procedure, and spyware operating traffic data. The random forest classification approach was used to verify this dataset and the suggested model. For data categorization, two approaches were used: binary-class classification and multi-class classification. In terms of precision, good results were obtained. The total average accuracy for binary-class classification was 79% and 77% for multi-class classification.