24th EANN 2023, 14 - 17 June 2023, León, Spain

BotDroid: Permission-based Android Botnet Detection Using Neural Networks

Saeed Seraj, Elias Pimenidis, Michalis Pavlidis, Stelios Kapetanakis, Marcello Trovati, Nikolaos Polatidis


  Android devices can now offer a wide range of services. They support a variety of applications, including those for banking, business, health, and entertainment. The popularity and functionality of Android devices, along with the open-source nature of the Android operating system, have made them a prime target for attackers. One of the most dangerous malwares is an Android botnet, which an attacker known as a botmaster can remotely control to launch destructive attacks. This paper investigates Android botnets by using static analysis to extract features from reverse-engineered applications. Furthermore, this article delivers a new dataset of Android apps, including botnet or benign, and an optimized multilayer perceptron neural network (MLP) for detecting botnets infected by malware based on the permissions of the apps. Experimental results show that the proposed methodology is both practical and effective while outperforming other standard classifiers in various evaluation metrics.  

*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.