Downloads: 0
Research Paper | Computer Science and Engineering | Volume 15 Issue 3, March 2026 | Pages: 702 - 710 | India
Blockchain-Secured Federated Learning with SMPC for Poisoning Attack Mitigation in Healthcare
Abstract: Federated Learning (FL) facilitates the joint model training between healthcare organizations without sharing the raw patient information. Despite the fact that the privacy can be better preserved with the aid of such a decentralized approach, it is still susceptible to poisoning attacks, where malicious participants will abuse and modify model updates to either deteriorate the global model's performance or introduce backdoors. To enhance privacy, integrity, and accountability, the current paper suggests a safe federated learning system that combines Secure Multi-Party Computation (SMPC) and a permissioned blockchain. SMPC secures local model updates through distributing encrypted shares such that it only exposes aggregated results such that individual contribution is not disclosed. The blockchain layer has records of identities of participants, update commitments, and model version histories that cannot be altered and provide tamper-proof auditing and trust management. The framework includes the use of strong aggregation, limited updates, safe validation, and weighting systems based on reputation to prevent the occurrence of poisoning behaviour. The analysis of experimental performance in a simulated healthcare cooperation setting has shown that there is substantial improvement in attack success rates, and at the same time, competitive model accuracy and computational efficiency have not been compromised. The suggested architecture has an effective and secure base on credible distributed healthcare artificial intelligence systems.
Keywords: Federated learning, secure multi-party computation, blockchain-based security, poisoning attack defense, robust model aggregation, healthcare data collaboration
How to Cite?: Manukonda Likhith Naveen Reddy, Jeffrin Hannah, "Blockchain-Secured Federated Learning with SMPC for Poisoning Attack Mitigation in Healthcare", Volume 15 Issue 3, March 2026, International Journal of Science and Research (IJSR), Pages: 702-710, https://www.ijsr.net/getabstract.php?paperid=SR26302215304, DOI: https://dx.dx.doi.org/10.21275/SR26302215304