🔧Challenges

Challenges Faced During Development

Building the decentralized system was not without its hurdles. Several key challenges arose, each requiring innovative solutions and extensive research to overcome:


1. Encryption Complexity and Synchronization

  • Challenge:

    • Implementing and integrating Pedersen Commitments, Paillier Cryptosystem, and AES Encryption was one of the most complex aspects of the project.

    • Each cryptographic scheme has unique properties, making it difficult to align them for seamless operation.

    • Ensuring compatibility and proper synchronization between these schemes while maintaining security and efficiency was a major technical hurdle.

  • Solution:

    • Focused on defining clear workflows for each encryption scheme, identifying their strengths, and deciding how to use them optimally.

    • Conducted rigorous testing to ensure the outputs from one encryption scheme could feed into another where required.


2. Maintaining Decentralization While Balancing On-Chain and Off-Chain Workloads

  • Challenge:

    • A core objective was to ensure the system remained truly decentralized, leveraging blockchains like Ethereum for aggregation and verification layers.

    • On-chain computations are inherently slower and significantly more expensive than off-chain alternatives, making it critical to decide which components should remain on-chain and which could be moved off-chain without compromising decentralization.

  • Solution:

    • Adopted a hybrid approach:

      • Kept critical actions such as aggregation, verification, and staking on-chain for transparency and trust.

      • Delegated intensive computations (e.g., local training) and storage operations (e.g., model files) to off-chain systems like Walrus.

    • Implemented decentralized checks and balances to ensure the off-chain operations were verifiable on-chain.


3. Lack of Established Resources and Research-Driven Implementation

  • Challenge:

    • There was no comprehensive guide or existing implementation for the problem we aimed to solve.

    • This required us to delve into academic research papers to understand the intricacies of cryptographic schemes, federated learning, and decentralized systems.

    • Translating theoretical concepts into practical, functional code was time-consuming and intellectually demanding.

  • Solution:

    • Dedicated significant time to reading and analyzing research papers, focusing on cryptographic primitives and their applications in secure federated learning.

    • Iteratively implemented, tested, and refined concepts derived from research until a reliable system was built.

Last updated