Submission to VIJ 2024-05-28
Keywords
- AndroRat, Android devices, Recurrent neural network, malware attack
Abstract
Android operating systems has gained increasing popularity in smart phones and other mobile intelligent terminals in recent years. Unpleasantly, the accumulation development and open nature of the platform has also attracted a vast number of malware developers. This paper presents a model for detecting and preventing AndroRat, a notorious Remote Administration Tool (RAT), on android devices. The research systematically pursued predefined objectives, leading to the development of a robust detection framework that utilizes the Outlier Technique. This framework significantly enhances the accuracy of AndroRat identification. Moreover, the dissertation introduces a preventive mechanism based on Recurrent Neural Network (RNN), fortifying Android phones against potential threats posed by RATs. By achieving the outlined objectives, this research makes a noteworthy contribution to the field of mobile device security. It demonstrates a practical application of machine learning techniques in mitigating evolving cybersecurity threats. A dataset was gathered for constructing the model. Object Oriented Analysis and Design (OOAD) was used as the research methodology and python was used as the programming language. The implementation of the Python Programming Language in this system not only ensured efficient execution but also established a versatile and accessible platform, laying the groundwork for future enhancements and adaptations. The comparative analysis conducted against existing systems highlights the effectiveness and innovation embedded in the proposed model, affirming its potential as a valuable addition to the realm of AndroRat detection and prevention with an accuracy of 99.99%. This dissertation not only addresses current security challenges but also establishes a foundation for continued advancements in mobile device security through the integration of cutting-edge technologies.