Integrating NLP and Ensemble Learning into Next-Generation Firewalls for Robust Malware Detection in Edge Computing

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Network Security

Background:

• Edge computing environments are vulnerable to sophisticated malware attacks that challenge traditional security measures.
• The increasing reliance on edge infrastructure necessitates advanced threat detection and mitigation strategies.
• Unstructured threat intelligence presents a rich data source for identifying novel attack vectors.

Purpose of the Study:

• To propose a novel framework integrating Natural Language Processing (NLP) and ensemble learning for malware detection in edge computing.
• To enhance the capabilities of Next-Generation Firewalls (NGFWs) for real-time threat identification and mitigation.
• To develop a scalable and adaptable security solution optimized for the constraints of edge environments.

Main Methods:

• Leveraging NLP techniques like TF-IDF vectorization to process unstructured threat intelligence (e.g., reports, logs).
• Implementing an ensemble learning model combining Random Forest (RF) and Logistic Regression (LR) using soft voting.
• Validating the model's robustness using ANOVA and confusion matrix analysis.

Main Results:

• Achieved 95% accuracy on a cyber threat intelligence dataset with synthetic data augmentation.
• Attained 98% accuracy on the CSE-CIC-IDS2018 dataset.
• Demonstrated low error rates and confirmed statistical robustness through validation analyses.

Conclusions:

• The proposed NLP and ensemble learning framework effectively detects and mitigates malware in edge computing environments.
• The system offers enhanced detection rates and adaptability, crucial for resource-constrained edge deployments.
• This approach provides a scalable defense layer for next-generation firewalls operating at the network edge.