BFL-SDWANTrust: Blockchain Federated-Learning-Enabled Trust Framework for Secure East-West Communication in...

Area of Science:

• Computer Networking
• Cybersecurity
• Machine Learning

Background:

• Software-Defined Wide-Area Networks (SD-WANs) are crucial for enterprise network reliability but are vulnerable to malicious nodes.
• Existing machine learning methods for detecting malicious nodes suffer from low accuracy and privacy concerns.
• Centralized SD-WAN architectures introduce single points of failure and performance bottlenecks.

Purpose of the Study:

• To propose a novel Blockchain Federated-Learning-Enabled Trust Framework (BFL-SDWANTrust) for secure east-west communication in multi-controller SD-WANs.
• To address the limitations of traditional SD-WANs by enhancing malicious node detection and ensuring data privacy.
• To eliminate reliance on third-party or centralized entities for network operations and validation.

Main Methods:

• Implemented a federated learning approach enabling local model training at edge nodes without central data aggregation.
• Developed a blockchain-based network to validate all network communications and malicious node detection transactions.
• Evaluated the BFL-SDWANTrust model on the InSDN dataset against benchmark malicious-node-detection models.

Main Results:

• BFL-SDWANTrust achieved superior performance compared to benchmark models across various metrics.
• The model demonstrated high accuracy (98.8%), precision (98.0%), recall (97.0%), and F1-score (97.7%).
• Achieved significantly reduced training (12 s) and testing (3.1 s) times.

Conclusions:

• The proposed BFL-SDWANTrust effectively enhances security and privacy in multi-controller SD-WANs.
• Federated learning and blockchain integration provide a robust, decentralized solution for malicious node detection.
• The framework offers improved generalization across heterogeneous SD-WAN environments while preserving network entity privacy.