TrustDroid: Enhancing Trust and Privacy for Android Mobile Phones by Preventing Riskware and SMS Malware

Android mobile phones are widely targeted by SMS Malware and Riskware attacks for stealing users' private information like user credentials, pictures, videos, banking credentials etc. Android users are not aware of the incoming link or message and unintentionally allow attackers to intrude on their mobile devices. The objective of this research paper is to enhance user confidence and privacy on Android smartphones. The CICMalDroid 2020 dataset provides the foundation for this research work by having a dataset on traffic patterns of riskware and SMS malware attacks on Android smartphones. In this research work, the methodology is proposed for early detection of riskware and SMS malware attacks by integrating a machine learning-trained chip at the hardware level. The dataset is trained with multiple machine learning algorithms to explore the effectiveness of appropriate algorithms for devising solutions for the detection of Android malware attacks. The results prove the efficiency of the random forest technique for the CICMalDroid dataset. This classifier has the best accuracy, recall, and
$F1$
score in comparison with other classifiers on the taken dataset. The Naive Bayes approach, on the other hand, is dependable and reduces the incidence of false positives. These results not only highlight the need to reinforce cybersecurity and regularly update software, but they also highlight the importance of machine learning in terms of Android device security.