SAutoIDS: A Semantic Autonomous Intrusion Detection System Based on Cellular Deep Learning and Ontology for Malware Detection in cloud computing

Abstract

Cloud computing (CC) is an online technology that has attracted the attention of many users and organizations today. Users send their requests through mobile to CC to perform a process. User requests are exposed to hacker attacks and malware risks. Penetration of malware in mobile devices causes loss of information or theft of mobile data. Today, various methods have been proposed to malware detection. In this paper, a semantic autonomous intrusion detection system (SAutoIDS) based on the ontology and cellular automata (CLA) and group method of data handling deep neural network (GMDH-DNN) is proposed to malware detection. The Semantic Multi-Level Approach (SMLA) processes of the data and transformed into semantic values based on a semantic level. The ontology method selects optimal features from malware data. Then the semantic data are divided into training (80%) and testing (20%). Training data are implemented to the GMDH-DNN for creating the model and CLA to optimize the GMDH model. Finally, testing data are entered into the optimized GMDH model and malwares are detected. We have used CICMalDroid2020 dataset to evaluate the SAutoIDS. By implementing the SAutoIDS, it was observed that the accuracy, precision, and recall improved by 21.96%, 22.41%, and 22.15%, compared to other methods.