Author(s) :   Pallavi Papankar¹, Nitin Mandaogade²

Abstract : Within the rapidly evolving realm of mobile technology, the imperative for robust security measures against Android malware is increasingly apparent. Traditional detection methods, grappling with the intricacies of contemporary threats, highlight the need for innovative strategies in malware identification and analysis. This review critically evaluates existing methodologies, spotlighting their limitations in accurately representing the intricate structure of Android applications (APKs) and effectively detecting malwares. These methodologies encounter challenges in addressing the dynamic and complex nature of APK components, leading to suboptimal detection rates and a lack of transparency in decision-making processes. Furthermore, scalability concerns persist as the Android application universe continues to expand. In response to these challenges, this review explores a pioneering framework based on Graph Neural Networks (GNNs) designed to revolutionize Android malware detection. The proposed model excels in feature extraction and the representation of APKs as graphs, capturing nuanced relationships among components such as permissions, API calls, and code segments. The core of this framework lies in its robust malware detection model, distinguishing between benign and malicious applications with unprecedented precision. Notably, the model prioritizes transparency and interpretability, offering insights into its decision-making process. Additionally, the review delves into the scalability and efficiency of the system, optimized for real-time analysis to handle the continuous influx of new Android applications. While empirical results are not included in this review, the discussion underscores the potential of the proposed architecture. This work sets a new standard in Android malware detection, providing a more accurate, transparent, and scalable solution. The review not only delivers a comprehensive overview of the current state of GNN-based approaches for Android malware detection but also lays the foundation for future research by proposing an architecture awaiting validation in subsequent studies.

Keywords: Graph Neural Networks, Android Malware Detection, Feature Extraction, Explainable AI, Scalable Security Systems.

DOI : 10.61161/ijarcsms.v12i3.2

Pages : 13-19

*Authors are invited to submit papers through E-mail at editor.ijarcsms@gmail.com