Thesis topic:

AI-Driven Blockchain-based Federated Learning for Edge Devices

• Supervisor: Mubashar Iqbal
  • contact: mubashar.iqbal@ut.ee
• This thesis aims to develop a framework that combines the decentralized nature of blockchain with federated learning to train AI models across edge devices while maintaining data privacy and transparency. In federated learning, AI models are trained locally on edge devices, allowing sensitive data to remain on the device rather than centralized in the cloud. This is particularly beneficial in sectors like healthcare and finance, where privacy and compliance are critical. Blockchain technology will be integrated to ensure secure, transparent, and immutable communication between edge devices, preventing tampering and ensuring trust in the data and model updates shared across the network. The framework will address key challenges such as model aggregation, secure data exchange, and resilience against malicious attacks on the learning process. Additionally, the blockchain's consensus mechanism will ensure that model updates are verified, preventing compromised edge devices from influencing the overall model. The thesis will also explore this hybrid approach's performance, scalability, and security.
• Reference paper: https://link.springer.com/article/10.1186/s12880-024-01279-4

<< back