Comparison of Convolutional Neural Network and Artificial Neural Network for Android Botnet Attack Detection

Abstract

Mobile devices, such as Androids, are now widely used. Androids are used for making phone calls, sending text messages, web browsing, social networking, and online banking transactions. The Android operating system's global popularity makes it a more appealing target for cyber criminals to gains access on Android device, to steal valuable data by installing an Android botnet attack. Thus, this research presents the Android botnet attack detection using deep learning algorithms, Convolutional Neural Network (CNN) and Artificial Neural Network (ANN). The experiment was carried out and tested on 1929 botnet dataset and 4873 benign applications using different categories of permission features. The research covers several performance metrics like accuracy, precision, recall, f1-score, true-positive and false-positive in identifying the best performance classifiers. At the end of the study, the  ANN classifier was identified to be best classifiers for Android Botnet attack detection with the highest detection accuracy 96.35% whereas the detection accuracy obtained by CNN is 95.44%.