RepuNet: A Reputation System for Mitigating Malicious Clients in DFL

2025-06-24

AI Summary

In decentralized federated learning (DFL), malicious nodes pose severe security threats—including model poisoning, delay attacks, and message flooding—yet systematic risk categorization and lightweight mitigation remain underexplored. This paper presents the first comprehensive taxonomy of DFL security risks and proposes a lightweight, adaptive dynamic reputation mechanism. Without relying on blockchain or auxiliary infrastructure, the mechanism evaluates node credibility in real time using multi-dimensional behavioral metrics: model similarity, parameter divergence, message latency, and communication volume; it then dynamically adjusts each node’s aggregation weight accordingly. The approach is topology-agnostic and robust to non-IID data distributions. Extensive experiments on MNIST and CIFAR-10 demonstrate strong resilience against diverse attack types and intensities, achieving F1 scores of 95.2% and 76.1%, respectively. The method significantly enhances both security guarantees and practical deployability of DFL systems.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Natural Language Processing: Safety and Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Decentralized Federated Learning (DFL) enables nodes to collaboratively train models without a central server, introducing new vulnerabilities since each node independently selects peers for model aggregation. Malicious nodes may exploit this autonomy by sending corrupted models (model poisoning), delaying model submissions (delay attack), or flooding the network with excessive messages, negatively affecting system performance. Existing solutions often depend on rigid configurations or additional infrastructures such as blockchain, leading to computational overhead, scalability issues, or limited adaptability. To overcome these limitations, this paper proposes RepuNet, a decentralized reputation system that categorizes threats in DFL and dynamically evaluates node behavior using metrics like model similarity, parameter changes, message latency, and communication volume. Nodes' influence in model aggregation is adjusted based on their reputation scores. RepuNet was integrated into the Nebula DFL platform and experimentally evaluated with MNIST and CIFAR-10 datasets under non-IID distributions, using federations of up to 25 nodes in both fully connected and random topologies. Different attack intensities, frequencies, and activation intervals were tested. Results demonstrated that RepuNet effectively detects and mitigates malicious behavior, achieving F1 scores above 95% for MNIST scenarios and approximately 76% for CIFAR-10 cases. These outcomes highlight RepuNet's adaptability, robustness, and practical potential for mitigating threats in decentralized federated learning environments.

Problem

• Mitigating malicious clients in decentralized federated learning
• Detecting model poisoning and delay attacks autonomously
• Reducing computational overhead without rigid configurations

Innovation

• Decentralized reputation system for DFL
• Dynamic node behavior evaluation metrics
• Reputation-based model aggregation adjustment