Enhancing Digital Supply Chain Security with AI: A Behavioral Analysis Approach Using CICIDS2017

Abstract

Digital supply chains are experiencing growing cybersecurity threats due to their dependency on digital networks which leads to operation disruptions together with sensitive data theft and financial damage. The security systems from earlier times use static rule-based detection methods yet remain unable to detect new security threats as they emerge. The proposed framework implements AI to monitor behavioral actions and detect anomalies in order to boost digital supply chain cybersecurity.

The researchers analyze intrusion detection through supervised and unsupervised learning methods by using data from the CIC-IDS2017 dataset along with its network traffic labels. Random Forest and XGBoost machine learning systems and deep learning models LSTMs and Autoencoders perform key roles as part of abnormal network pattern assessment. The research checks unsupervised anomaly detection approaches through their anomaly detection methods of clustering and isolation forests which focus on detecting threats without labeled threats. Testing of the proposed models occurs through evaluation against essential performance metrics which balance detection precision with false alarm rates using accuracy along with precision and recall and F1-score and ROC-AUC measures.

The field experimental work shows that artificial intelligence-based cybersecurity detection methods have proved superior to traditional security protocols. AI technology will merge with blockchain systems for developing real-time adaptive cybersecurity frameworks as the study investigates these possibilities. The research results will help strengthen supply chain security and create a basis for deploying AI intrusion detection technology in sectors including healthcare and energy infrastructure