Malware Attack Detection using Machine Learning Methods for IoT Smart Devices

Abstract

The malware attacks are targeting IoT devices as the rapid development of these devices. The limited resource of IoT devices is attracting malware developers. The strong security mechanisms cannot be deployed on these devices because of their computational capabilities. Therefore, there are malicious attacks challenging these devices, especially botnet attacks. After infection to these devices, they tried to attack the victim user by launching the distributed denial of service (DDoS). Although machine learning methodologies can support to detect these attacks, their heavyweight processing is challenging to implement the prompt response to the attack actions. Therefore, this paper intends to reduce the processing time by using the information-gain feature selection method for implementing the malware attack detection system with the CART learning algorithm, and its results are compared the performance with Naïve Bayes. The experiment results indicate that the proposed methodology is effective in detecting malware attacks with up to 100% accuracy.