A Robust Framework for MADS Based on DL Techniques on the IoT

Abstract

Day after day, new types of malware are appearing, renewing, and continuously developing, which makes it difficult to identify and stop them. Some attackers exploit artificial intelligence (AI) to create renewable malware with different signatures that are difficult to detect. Therefore, the performance of the traditional malware detection systems (MDS) and protection mechanisms were weakened so the malware can easily penetrate them. This poses a great risk to security in the internet of things (IoT) environment, which is interconnected and has big and continuous data. Penetrating any of the things in the IoT environment leads to a penetration of the entire IoT network and control different devices on it. Also, the penetration of the IoT environment leads to a violation of users’ privacy, and this may result in many risks, such as obtaining and stealing the user’s credit card information or theft of identity. Therefore, it is necessary to propose a robust framework for a MDS based on DL that has a high ability to detect renewable malware and propose malware Anomaly detection systems (MADS) work as a human mind to solve the problem of security in IoT environments. RoMADS model achieves high results: 99.038% for Accuracy, 99.997% for Detection rate. The experiment results overcome eighteen models of the previous research works related to this field, which proved the effectiveness of RoMADS framework for detecting malware in IoT.