Multi-Level Intrusion Detection System in Cloud Computing
Stephen Alaba Mogaji · ABUAD International Journal of Natural and Applied Sciences · 2025
An intrusion detection system (IDS) is essential for protecting private data, maintaining system integrity, and ensuring network security. This research focused on the development of a multi-level IDS using cloud computing technologies to enhance network security and employs a multi-layered strategy for intrusion detection, encompassing different levels for the scrutiny of network traffic, system behavior, and potential security risks. Cloud computing infrastructure forms the basis for deploying and expanding the IDS effectively. Anomaly-based intrusion detection systems have poor precision and recall, especially for unidentified attack types and this is seen as a research gap. Alternative Fuzzy C-Means Clustering (AFCM) was utilized in this work to group the training data into homogenous training subsets, train various Artificial Neural Networks (ANN) using those subsets, and then aggregate the results. The neural network was trained using the KDD Cup ‘99 dataset and was tested using real-time internet traffic. The SYN flood, ICMP flood, and Normal activity were used as test cases for the attack and normal activity, the results show that the SYN has a 98.9% true positive rate and 1.1% false negative rate of the 10000number of connections, ICMP has 99.9% true positive rate and 0.1% false negative rate of the 10000 number of connections and the normal activity has 88.09% true positive rate and 11.74% false negative rate of the 10000 number of connections. This is an improvement over other common anomaly-based intrusion detection systems.