Unauthorised access constitutes a primary security threat in decentralised networks, where the absence of a single control point multiplies attack surfaces and complicates timely detection. This paper presents a rigorous performance evaluation of a Random Forest classifier engineered to distinguish authorised from unauthorised network activity on the basis of five readily extractable traffic features: packet rate, bandwidth usage, latency, failed login attempts and connection duration. An Isolation Forest component operates in parallel as an unsupervised anomaly detector, supplying coverage for previously unseen (zero-day) behavioural patterns. Experiments conducted on labelled network traffic yield an accuracy of 96.80%, precision of 95.40%, recall of 94.20% and F1-score of 94.80%. By comparison, a conventional signature-based intrusion detection baseline attains only 82.40%, 80.10%, 78.60% and 79.30% on the same metrics. The associated confusion matrix reveals low false-positive and false negative counts (18 and 14, respectively, on a balanced test set of approximately one thousand samples). These results establish Random Forest as a practical, interpretable and computationally efficient solution for real-time unauthorised user classification in resource-constrained decentralised settings, while Isolation Forest further strengthens resilience against novel attacks.

REFERENCES

[1]. Farhan, M. (2025). Transformer-based architectures for network intrusion detection: Advances and challenges. IEEE Access, 13, 45678–45690.

[2]. Lu, C., Cao, Y., & Wang, Z. (2024). Research on intrusion detection based on an enhanced random forest algorithm. Applied Sciences, 14(2), 714.

[3]. Moustafa, N., & Slay, J. (2018). UNSW-NB15: A comprehensive data set for network intrusion detection systems. Military Communications and Information Systems Conference.

[4]. Nhlapo, S. J., & Nkongolo, M. N. (2024). Zero-day attack and ransomware detection. arXiv. https://doi.org/10.48550/arXiv.2408.05244.

[5]. Pramilarani, K., Rajkumar, N., et al. (2024). Cost-based Random Forest classifier for intrusion detection system. Applied Soft Computing.

[6]. Sharafaldin, I., Lashkari, A. H., & Ghorbani, A. A. (2018). Toward generating a new intrusion detection dataset and intrusion traffic characterization. ICISSP, 108–116.

[7]. Sommer, R., & Paxson, V. (2010). Outside the closed world: On using machine learning for network intrusion detection. IEEE Symposium on Security and Privacy.

[8].  Thakkar, A., & Lohiya, R. (2020). A survey on intrusion detection system: Feature selection, model, performance measures, application perspective, challenges, and future research directions. Artificial Intelligence Review, 53(7), 5111 5152.

[9]. Zhang, H. (2025). Semi-supervised learning methods for intrusion detection in large-scale networks. Expert Systems with Applications, 237, 121531.

[10]. Canadian Institute for Cybersecurity. (2017). CICIDS2017 dataset. University of New Brunswick.