Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AutoDFL: A Scalable and Automated Reputation-Aware Decentralized Federated Learning

Meryem Malak Dif, Mouhamed Amine Bouchiha, Mourad Rabah, Yacine Ghamri-Doudane

TL;DR

AutoDFL tackles scalability and on-chain cost concerns in blockchain-based decentralized federated learning by integrating Layer-2 zk-Rollups with an automated reputation framework and a DON-enabled data flow. It defines a dual-layer blockchain, IPFS off-chain storage, and a reputation model that blends objective performance with subjective trust signals, all orchestrated through smart contracts. The proof-of-concept demonstrates throughput exceeding $3000$ TPS with up to a $20\times$ gas reduction, validating the approach's practicality for large-scale, reputation-aware FL. This work offers a scalable, transparent, and incentive-compatible framework for secure, decentralized collaborative learning in real-world deployments.

Abstract

Blockchained federated learning (BFL) combines the concepts of federated learning and blockchain technology to enhance privacy, security, and transparency in collaborative machine learning models. However, implementing BFL frameworks poses challenges in terms of scalability and cost-effectiveness. Reputation-aware BFL poses even more challenges, as blockchain validators are tasked with processing federated learning transactions along with the transactions that evaluate FL tasks and aggregate reputations. This leads to faster blockchain congestion and performance degradation. To improve BFL efficiency while increasing scalability and reducing on-chain reputation management costs, this paper proposes AutoDFL, a scalable and automated reputation-aware decentralized federated learning framework. AutoDFL leverages zk-Rollups as a Layer-2 scaling solution to boost the performance while maintaining the same level of security as the underlying Layer-1 blockchain. Moreover, AutoDFL introduces an automated and fair reputation model designed to incentivize federated learning actors. We develop a proof of concept for our framework for an accurate evaluation. Tested with various custom workloads, AutoDFL reaches an average throughput of over 3000 TPS with a gas reduction of up to 20X.

AutoDFL: A Scalable and Automated Reputation-Aware Decentralized Federated Learning

TL;DR

AutoDFL tackles scalability and on-chain cost concerns in blockchain-based decentralized federated learning by integrating Layer-2 zk-Rollups with an automated reputation framework and a DON-enabled data flow. It defines a dual-layer blockchain, IPFS off-chain storage, and a reputation model that blends objective performance with subjective trust signals, all orchestrated through smart contracts. The proof-of-concept demonstrates throughput exceeding TPS with up to a gas reduction, validating the approach's practicality for large-scale, reputation-aware FL. This work offers a scalable, transparent, and incentive-compatible framework for secure, decentralized collaborative learning in real-world deployments.

Abstract

Blockchained federated learning (BFL) combines the concepts of federated learning and blockchain technology to enhance privacy, security, and transparency in collaborative machine learning models. However, implementing BFL frameworks poses challenges in terms of scalability and cost-effectiveness. Reputation-aware BFL poses even more challenges, as blockchain validators are tasked with processing federated learning transactions along with the transactions that evaluate FL tasks and aggregate reputations. This leads to faster blockchain congestion and performance degradation. To improve BFL efficiency while increasing scalability and reducing on-chain reputation management costs, this paper proposes AutoDFL, a scalable and automated reputation-aware decentralized federated learning framework. AutoDFL leverages zk-Rollups as a Layer-2 scaling solution to boost the performance while maintaining the same level of security as the underlying Layer-1 blockchain. Moreover, AutoDFL introduces an automated and fair reputation model designed to incentivize federated learning actors. We develop a proof of concept for our framework for an accurate evaluation. Tested with various custom workloads, AutoDFL reaches an average throughput of over 3000 TPS with a gas reduction of up to 20X.
Paper Structure (24 sections, 9 equations, 5 figures, 2 tables)

This paper contains 24 sections, 9 equations, 5 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. Proposed Framework
  4. System Model
  5. Threats Model
  6. System Architecture
  7. Task Publishers Layer
  8. Federated Learning Layer
  9. Dual-Layered Blockchain with zk-Rollups
  10. Off-Chain Storage
  11. Decentralized Oracles Network (DON)
  12. System Workflow
  13. Reputation Model
  14. Objective Reputation
  15. Subjective Reputation
  16. ...and 9 more sections

Figures (5)

  • Figure 1: System Architecture. ASC refers to the Access Smart Contract, which enforces role-based control and regulates permissions, granting access solely to authorized users.
  • Figure 2: Dual-Layered Blockchain Design: L1 is an Ethereum Virtual Machine (EVM)-based blockchain, L2 is powered by zk-Rollups.
  • Figure 3: Reputation Dynamics of Profiles With Varying Behaviors: Good, Malicious, and Lazy Participation.
  • Figure 4: L1 Throughput and latency comparison under different workload types.
  • Figure 5: Average Throughput Comparison: Single-Layered BFL vs AutoDFL.