Neural Network Based Android Malware Detection with Different IP Coding Methods

Abstract

Due to the COVID-19 epidemic that has affected the whole world, internet use has increased more than in previous years. Almost all operations and transactions are done over the internet, especially with the use of cellular phones and tablet PCs. This growth results in many security deficits that need to be solved by security admins and end users. Malicious software (malware) is generally preferred for attacking the computer systems and recently for cellular phones. As a mobile operating system, Android is the main player of this sector with about 72% market share worldwide. Therefore, malware attacks especially target these devices, for reaching the maximum number of victims. The situation is getting more and more devastating with around 12,000 new Android malware attacks every day. This is one critical problem that needed to be solved by setting up an android malware detection system. Machine learning algorithms are frequently preferred in data mining-based security applications which contain lots of features in datasets. Artificial Neural networks are one of the mostly preferred learning models for training the system. Therefore, in this paper, it is aimed to implement a neural network based android malware detection system by using an up-to-date dataset presented by the Cyber Security Institute of Canada as CICMalDroid2017. Ip Addresses are one of the features in this dataset, and we focus on two different IP coding methods, as IP Splitting to Four Numbers, IP Transform to integer number, and no IP Address. In experimental study we reached a good level of accuracy rate as 98.4% by splitting an IP address to four numbers.