Abstract:Important information needs to be sent over the Internet safely. Real-time data is mixed with harmful information, lowering the quality of communication and the system's overall performance. A network intruder detection system is software that examines all incoming and outgoing network packets to detect malicious events. Federated learning (FL) is a way to put artificial intelligence at the cutting edge. It is a way to solve problems that is not centralized and lets people learn from significant amounts of data. Deep learning (DL) methods have often been used to find harmful data in host intrusion detection systems (HIDS) that look for unusual behavior. The FL architecture allows multiple users to train a global model while respecting the privacy of each user's data, making DL-based methods more useful. But there has yet to be a complete analysis of how well FL-based HIDSs protect against known privacy threats with the already in-place defenses. To solve this problem, we offer two privacy assessment measures for FL-based HIDSs, including a privacy score that rates how close the original and restored traffic attributes are. The CICIDS2017 dataset, which includes several attacks from the present day, was used to make the real-time model. In addition, an adaptive threshold-correlation algorithm (ATCA) is presented to enhance detection accuracy by dynamically adjusting threshold values according to traffic patterns and intrusion behaviors. The FL-HIDS framework was created and tested using a realistic network dataset. Experiment results show that the suggested technique outperforms existing intrusion detection systems regarding detection precision and scalability. The federated learning strategy effectively leverages the collective intelligence of network devices, enabling continuous learning and adaptation to emergent attack strategies. Furthermore, the adaptive threshold technique considerably reduces the rate of false positive and false negative detection, boosting the intrusion detection system's overall effectiveness. The proposed architecture solves previous centralized solutions' shortcomings by providing a scalable, privacy-preserving method for defending network environments against expanding invasion threats.

Privacy-preserving Few-shot Traffic Detection against Advanced Persistent Threats via Federated Meta Learning

A Method of Few-Shot Network Intrusion Detection Based on Meta-Learning Framework

Backdoor Attacks and Defenses in Federated Learning: State-of-the-Art, Taxonomy, and Future Directions

A Federated Learning Approach for Multi-stage Threat Analysis in Advanced Persistent Threat Campaigns

Detecting Advanced Persistent Threat in Edge Computing via Federated Learning

XFedHunter: An Explainable Federated Learning Framework for Advanced Persistent Threat Detection in SDN

Few-shot Multi-domain Knowledge Rearming for Context-aware Defence against Advanced Persistent Threats

FL-PTD: A Privacy Preserving Defense Strategy Against Poisoning Attacks in Federated Learning.

Enabling Privacy-Preserving Cyber Threat Detection with Federated Learning

FedDef: Defense Against Gradient Leakage in Federated Learning-Based Network Intrusion Detection Systems

Predicting the APT for Cyber Situation Comprehension in 5G-Enabled IoT Scenarios Based on Differentially Private Federated Learning

Poisoning Attacks on Federated Learning-based Wireless Traffic Prediction

Model-agnostic generation-enhanced technology for few-shot intrusion detection

Meta-Fed IDS: Meta-Learning and Federated Learning Based Fog-Cloud Approach to Detect Known and Zero-Day Cyber Attacks in IoMT Networks

An Evolutionary Computation-Based Federated Learning for Host Intrusion Detection in Real-Time Traffic Analysis

Fed-LSAE: Thwarting Poisoning Attacks against Federated Cyber Threat Detection System via Autoencoder-based Latent Space Inspection

Decentralized federated meta‐learning framework for few‐shot multitask learning

Mitigating Backdoors in Federated Learning with FLD

3DFFL: privacy-preserving Federated Few-Shot Learning for 3D point clouds in autonomous vehicles

APFed: Adaptive personalized federated learning for intrusion detection in maritime meteorological sensor networks

NFHP-RN: A Method of Few-Shot Network Attack Detection Based on the Network Flow Holographic Picture-ResNet