Scholarlink

<p>With the rapid evolution of Fifth Generation (5G) networks and the surge of digital healthcare data, ensuring privacy, trust, and low-latency learning across medical institutions has become imperative. This paper presents <i>BlockFed</i>, a blockchain-enabled federated learning (FL) framework that safeguards patient health records (PHR) while enabling collaborative intelligence among healthcare entities. In <i>BlockFed</i>, model training occurs locally at hospitals or diagnostic centers, and only encrypted model updates secured through additive homomorphic encryption are shared. The blockchain layer records update hashes via smart contracts, ensuring immutability, verifiability, and access control, while the InterPlanetary File System (IPFS) offloads encrypted weights to minimize on-chain storage overhead. To maintain real-time responsiveness, a Practical Byzantine Fault Tolerance (PBFT) consensus ensures sub-second finality and deterministic security in 5G-assisted environments. Experimental results on brain MRI datasets demonstrate that <i>BlockFed</i> achieves 95.1% classification accuracy, comparable to centralized models (96.3%), with only a marginal 1.2% trade-off in accuracy. The framework reduces communication cost by 84.5% and achieves a total computation delay of approximately 309 ms per update transaction, validating its efficiency for low-latency healthcare scenarios. <i>BlockFed</i> thus offers a scalable, verifiable, and privacy-preserving foundation for next-generation 5G healthcare ecosystems.</p>

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

BlockFed: Blockchain-based privacy preserving federated learning for 5G-assisted healthcare ecosystems

  • 期刊论文

  • 发表日期：

    2026年7月23日

查看全文
  • Ashwin Verma,
  • Sunil Pathak,
  • Pronaya Bhattacharya

摘要

With the rapid evolution of Fifth Generation (5G) networks and the surge of digital healthcare data, ensuring privacy, trust, and low-latency learning across medical institutions has become imperative. This paper presents BlockFed, a blockchain-enabled federated learning (FL) framework that safeguards patient health records (PHR) while enabling collaborative intelligence among healthcare entities. In BlockFed, model training occurs locally at hospitals or diagnostic centers, and only encrypted model updates secured through additive homomorphic encryption are shared. The blockchain layer records update hashes via smart contracts, ensuring immutability, verifiability, and access control, while the InterPlanetary File System (IPFS) offloads encrypted weights to minimize on-chain storage overhead. To maintain real-time responsiveness, a Practical Byzantine Fault Tolerance (PBFT) consensus ensures sub-second finality and deterministic security in 5G-assisted environments. Experimental results on brain MRI datasets demonstrate that BlockFed achieves 95.1% classification accuracy, comparable to centralized models (96.3%), with only a marginal 1.2% trade-off in accuracy. The framework reduces communication cost by 84.5% and achieves a total computation delay of approximately 309 ms per update transaction, validating its efficiency for low-latency healthcare scenarios. BlockFed thus offers a scalable, verifiable, and privacy-preserving foundation for next-generation 5G healthcare ecosystems.