2025/10/17 by Jothy Chandrikha Raveendran, J Edwin, Anju Ajith Jaya · 1 voice
Computer Science · #Network Security and Intrusion Detection #Advanced Malware Detection Techniques #Anomaly Detection Techniques and Applications
paper · doi:10.70389/pjs.100137
openalex publication_date 2025/10/17 · openalex created_date 2025/10/28 · openalex updated_date 2026/07/20
As digitalization has accelerated globally, intrusion becomes most vulnerable due to which Intrusion Detection Systems (IDS) gains more attention in cyber security for detecting unauthorized access and malicious activities within networks and systems. Recent advancements in technology paves way to integrate these IDS with new technologies like Machine Learning (ML), Deep Learning (DL), Blockchain (BC) and so on for addressing the growing sophistication of cyber threats. Traditional IDS actually experience high false positives, poor adaptability, and scalability problems owing to their dependencies on static rules and centralized designs. They also have no means of safeguarding data integrity and identity authentication, leaving them exposed to log tampering and spoofing attacks. Our proposed ensemble ML-based IDS with Blockchain integration provides tamper-resistant logs, safe identity management, and decentralized scalability. ML provides promising solutions through data-driven algorithms, though it requires large datasets and can be vulnerable to adversarial attacks. Blockchain offers a decentralized, tamper-proof ledger, enhancing IDS security and reliability but faces hurdles like computational overhead and scalability. This paper proposes a stacked ensembled IDS framework that synergizes ensemble machine learning algorithm which is a combination of Random Forest and Gradient Boost algorithms, which is integrated with the blockchain technology for decentralized alert sharing. Both the algorithms are highly effective for detecting intrusions, where the former offers a strong robustness and the later by repeatedly correcting the error improves the accuracy of prediction. This hybrid IDS framework aims to improve detection accuracy, reduce false positives, and ensure data integrity, contributing to more secure, adaptable, and efficient IDS solutions.