LAUSR

Abstract Green and smart cities deliver services to their residents using mobile applications that make daily life more convenient. The privacy and security of these applications are significant in providing sustainable services in a green city. The software cloning is a severe threat which may breach the security and privacy of android applications. A centrally controlled and automated screening system across multiple app stores is inevitable to prevent the release of copyrighted or cloned copies of these apps. In this paper, we proposed IoT-enabled green city architecture for clone detection in android markets using a deep learning approach. First, the proposed system obtained an original APK file together with potential candidate cloned APKs via the cloud network. For each subject software, the system uses an APK Extractor tool to retrieve Dalvik Executable (DEX) files. The Jdex decompiler is utilized to retrieve Java source files through Dalvik Executables. Second, the AST features are extracted using ANother Tool for Language Recognition (ANTLR) parser. Third, the linear features are mined from these hierarchical structures, and Term Frequency Inverse Document Frequency (TFIDF) is applied to estimate the significance of each feature. Finally, the deep learning model is configured to detect cloned apps. The deep learning model is fine-tuned to get better accuracy. The proposed approach is analyzed on five different cloned applications collected from different android markets. The main objective of this system is to avoid the release of pirated apps with various pirated labels in multiple app markets.