Detection of clone scammers in Android markets using IoT-based edge computing

Pirated application developers find an alternate way to publish pirated versions of the same Android mobile applications (apps) on different Android markets. Therefore, a centralized, automated scrutiny system among multiple app stores is inevitable to prevent publishing pirated or cloned version of these Android applications. In this paper, we proposed an Android clone detection system for Internet of things (IoT) (Droid-IoT) devices. First, the proposed system receives an original Android application package (APK) file along with possible candidate cloned APKs over the cloud network. The system uses an apkExtractor tool to extract Dalvik Executable (DEX) files for each subject program. The Jdex decompiler is used to extract Java source files from DEXs. Then, the bag-of-word model is used to extract tokenized features from source files. Further, the weighting filters are used to zoom the importance of each token. Moreover, Synthetic Minority Oversampling is applied to retrieve balanced features for better training of data. Finally, TensorFlow with Keras deep learning model is designed to predict clones in Android applications. The experimental results have shown that Droid-IoT can successfully detect cloned apps with an accuracy of up to 96%. The primary purpose of this system is to prevent the publishing of pirated apps among different app stores under different pirated names.