Advancements in Android Malware Detection: Comprehensive Technical Analysis of Deep Learning Techniques

Nowadays, Android smartphones have occupied about 72% of the mobile market worldwide. This huge number of Android users attracts large-scale malware attacks due to the open-source nature of Android operating systems. Android app installation with extensive permission requests granted by smartphone users enables malware to access device functionality and sensitive data. Due to the limitations of traditional signature-based and heuristic-based malware detection methods, it is very important to protect against these attacks using evolving deep learning techniques. This review study covers and summarizes the latest academic papers over the last 12 years on malware detection architectures and categorizes them as per the use of dynamic analysis, static analysis, and hybrid analysis. This review will assist researchers in outlining future research directions in Android malware detection using deep learning techniques such as CNNs, LSTMs, Federated Learning, Graph Neural Networks, and GANs. Furthermore, the review identifies critical research gaps and proposes directions for future investigation to develop more robust, interpretable, and privacy-preserving malware detection systems.