A Hybrid LSTM Autoencoder and Isolation Forest Framework for Anomaly-Based Ransomware Detection in Network Traffic
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
In this study, a novel anomaly detection-based hybrid model is realized for the prediction of ransomware attacks by monitoring network traffic. The signature or rule-based traditional approach to combating polymorphic ransomware and zero-day attack is under severe strain, especially in the presence of encrypted network traffic, that requires using more adaptive and flexible approaches that analyse behaviours. The proposed framework exploits temporal patterns of traffic by LSTM autoencoder and identifies statistically anomalous values with Isolation Forest. The outputs from the two models are then combined by a weighted merge mechanism to a final anomaly score to determine if traffic is benign or malicious. The proposed model was tested on an enhanced version of UGRansome_Dataset_2024 dataset having 111,904 records in unbalanced form, and balanced by SMOTE-ENN technique up to 149,197 records containing benign and malicious traffic. The simulation results demonstrate the superiority of the proposed model with 99.66% accuracy, 99.78% precision, 99.85% recall, 99.81% F1-score. These results demonstrate that the combined approach can effectively reduce false alarms, while the ability to detect attacks is significantly improved, and therefore can be used in practice networked environments.
Fields of Study & Subjects
Advanced Malware Detection Techniques
Network Security and Intrusion Detection
Internet Traffic Analysis and Secure E-voting
Publication details
- Journal / source
- International Journal of Advances in Signal and Image Sciences
- Volume
- 12
- Issue
- 3s
- Pages
- 1394–1402
- Publication date
- 9 March 2026
- Language
- EN