FetFIDS: A Feature Embedding Attention based Federated Network Intrusion Detection Algorithm

2025-08-12

AI Summary

To address privacy preservation, communication overhead, and local model performance in Transformer-based network intrusion detection systems (IDS) deployed at the edge, this paper proposes FetFIDS—the first federated IDS framework that replaces conventional positional embeddings with feature embeddings. By explicitly modeling salient semantic features of network traffic, FetFIDS enhances the Transformer’s capacity to capture discriminative patterns, while synergistically optimizing local model updates via a lightweight federated learning protocol. Evaluated on NSL-KDD, CIC-IDS2017, and other benchmark datasets, FetFIDS consistently outperforms state-of-the-art methods: it achieves an average 3.2% improvement in detection accuracy and a 4.7% gain in F1-score, while reducing client-side parameter upload volume by 68%. The framework thus delivers superior detection performance, strong privacy guarantees, and efficient adaptability to resource-constrained edge environments.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Intrusion Detection Systems (IDS) have an increasingly important role in preventing exploitation of network vulnerabilities by malicious actors. Recent deep learning based developments have resulted in significant improvements in the performance of IDS systems. In this paper, we present FetFIDS, where we explore the employment of feature embedding instead of positional embedding to improve intrusion detection performance of a transformer based deep learning system. Our model is developed with the aim of deployments in edge learning scenarios, where federated learning over multiple communication rounds can ensure both privacy and localized performance improvements. FetFIDS outperforms multiple state-of-the-art intrusion detection systems in a federated environment and demonstrates a high degree of suitability to federated learning. The code for this work can be found at https://github.com/ghosh64/fetfids.

Problem

• Improving intrusion detection with feature embedding transformers
• Enhancing federated learning for network security privacy
• Optimizing edge deployment for localized performance gains

Innovation

• Feature embedding replaces positional embedding in transformers
• Federated learning ensures privacy across multiple communication rounds
• Designed for edge deployment with localized performance improvements