OHMN: Android Malware Detection Method Using Opcode Highway Memory Network

Abstract Malware developed for Android devices has recently showed a preference for rapid replication, which speeds up the process of stealing sensitive data. By reusing code, hackers introduce potentially harmful modifications. In this study, we suggest using a deep learning-based network called the "Opcode highway memory network" (OHMN), which may identify versions of malware that are related to one another without requiring the analyst to have any specialized understanding in mathematics or methodology. Truth interference fuzzy clustering is a way to find features using software birth marking of malcode sequences, and sequence patch normalization may be used to standardize the input data. Malware samples are clustered, and that information is sent into the OHMN process that looks for different strains of malware. The Android malware detection system has been shown to operate well via both observation and experimentation. General testing was performed on the Mal Radar dataset, an android malware dataset obtained from the zenodo under python environment. As shown by the simulation results, the proposed approach outperformed the state-of-the-art algorithm.