An Intelligent Malware Detection Model For Android Mobile Devices

Abstract

Android devices have experienced an immense popularity over the last few years. And this growth has exposed these devices to an increasing number of security threats. Despite the variety of peripheral protection mechanisms such as Bouncer, authentication and access control cannot provide integral protection against intrusions. Thus, the need for a more sophisticated security controls such as anomaly detection systems is necessary. Whilst much work has been devoted to mobile device intrusion detection systems, research on Android anomaly‐based has been limited leaving several problems unsolved. Such as getting an error free detection technique or at least reducing the error to the barest minimum. Motivated by this fact, the researcher focused on anomaly technique of detecting Botnet on Android mobile devices.  An open source malware database called Kaggle.com which is said to have a high malware detecting power was used. A cross‐evaluation of three Machine Learning algorithms (i.e. Support Vector Machine (SVM), Random Forest (RF), and K-Nearest Neighbour (KNN. To check which of them detect with the highest accuracy and low False positive rate were considered. At the end the results of the three machine Learning models were then compared to know the best classifier for the model, and Random Forest came out as the best classifier with 99.99% accuracy and 0.013% false positive rate.