Detecting Spyware in Android Devices Using Random Forest

Abstract

Even the most cautious user cannot guarantee that avoidance alone will completely protect them from cell phone spyware because mobility plays an increasingly important role in both business and leisure. Advanced Android spyware has become more prevalent as a result of the recent and obvious rise in mobile devices. Currently available Android spyware and other malicious software are frequently camouflaged and buried among the many useful Android apps. These sophisticated spyware skulks in the user-trusted third-party program market may jeopardize the security of the user’s smart device, resulting in financial or personal loss. The spyware makers also distributed their programs under names that were obscured to entice users to install them on their mobile devices. Applications run in the background throughout this procedure and display adverts to prevent users from detecting spyware. Correctly classifying spyware using traditional detection approaches takes a lot of time and human work. Deep learning (DL) based spyware detection and classification approaches have recently been created to address this problem. This study’s objective is to provide a revolutionary methodology that, can successfully detect Android spyware even in the face of tricky and evasive attacks. The main advantage was the utilization of recently specialized datasets from the real-world environment to create more reliable and adaptive training and testing materials, which were then combined with different algorithms to create the most accurate and time-efficient model feasibly. The best outcomes were obtained with 92% accuracy from the Random Forest method. The dataset is thought to offer greater potential and good outcomes, although the model is not assured against recently developing spyware technologies and obfuscation.