Data exploration algorithms in anomaly detection in communication protocols

Cybersecurity in modern communication networks is of paramount importance, particularly in critical infrastructure sectors. Anomaly detection in communication protocols is a key component in identifying and mitigating cyber threats. This study explores data-centric approaches for anomaly detection using machine learning algorithms. We evaluate the effectiveness of ensemble models incorporating Isolation Forest, XGBoost, and Autoencoders to reduce false positives while maintaining high accuracy. Our methodology involves training on both labeled and unlabeled datasets, including NSL-KDD and CIC-IDS2017, to simulate real-world attack scenarios. Experimental results demonstrate that the proposed ensemble learning approach enhances detection performance, offering a balanced trade-off between precision and false alarm reduction. These findings contribute to the development of robust and scalable intrusion detection systems suitable for deployment