Scholarlink

The rapid proliferation of distributed systems, Internet of Things (IoT) devices, and cloud-based services has introduced unprecedented challenges to data security and privacy. Traditional centralized approaches to data protection often struggle to balance security, scalability, and efficiency, particularly when sensitive data is involved. This paper proposes a federated learning-based framework to enhance data security and privacy across distributed environments. By enabling local model training on edge devices without transmitting raw data to centralized servers, the framework significantly reduces privacy risks and potential attack surfaces. We present the architecture, key components, and security mechanisms embedded in the framework, including secure aggregation, differential privacy, and homomorphic encryption. Extensive experiments were conducted on benchmark datasets to evaluate the framework’s performance in terms of accuracy, communication overhead, and resilience against adversarial attacks. The results demonstrate that the proposed approach not only preserves privacy but also achieves competitive predictive performance compared to traditional centralized models. This work contributes a scalable and practical solution to address the evolving demands of secure and privacy-preserving machine learning in real-world distributed applications.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Data Security and Privacy Using Federated Learning: A Scalable Framework for Distributed Systems

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ammar Odeh,
  • Walid Salameh,
  • Tareq Alhajahjeh,
  • Francisco Navarro,
  • Aladdin Ayesh,
  • Mohammad Abu Karim

摘要

The rapid proliferation of distributed systems, Internet of Things (IoT) devices, and cloud-based services has introduced unprecedented challenges to data security and privacy. Traditional centralized approaches to data protection often struggle to balance security, scalability, and efficiency, particularly when sensitive data is involved. This paper proposes a federated learning-based framework to enhance data security and privacy across distributed environments. By enabling local model training on edge devices without transmitting raw data to centralized servers, the framework significantly reduces privacy risks and potential attack surfaces. We present the architecture, key components, and security mechanisms embedded in the framework, including secure aggregation, differential privacy, and homomorphic encryption. Extensive experiments were conducted on benchmark datasets to evaluate the framework’s performance in terms of accuracy, communication overhead, and resilience against adversarial attacks. The results demonstrate that the proposed approach not only preserves privacy but also achieves competitive predictive performance compared to traditional centralized models. This work contributes a scalable and practical solution to address the evolving demands of secure and privacy-preserving machine learning in real-world distributed applications.