Table 34 Limitations regarding computational demands of Blockchain + federated learning- observed/implied limitations.

Computational overhead

Blockchain components like SHA-256 hashing, proof-of-authority (PoA) consensus, and transaction validation require additional computation.

May slow down model update cycles, especially in resource-constrained hospitals or edge devices.

Latency in model aggregation

The time taken to encrypt, verify, and record updates on-chain can introduce delays in federated learning rounds.

Delays can reduce responsiveness in real-time diagnostic systems.

Energy consumption

Continuous model updates and blockchain synchronization may lead to higher power consumption, particularly on devices not optimized for such tasks.

Not ideal for battery-operated or embedded medical systems (e.g., portable CT/MRI units).

Infrastructure requirements

Deploying private Ethereum networks for PoA requires networking, node deployment, and secure key management.

Smaller healthcare facilities may lack the technical infrastructure to support this.

Scalability issues

As the number of participating institutions grows, blockchain synchronization becomes more complex and bandwidth-intensive.

Could hinder national/global-scale collaborative learning deployments unless optimized.