Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FWeb3: A Practical Incentive-Aware Federated Learning Framework

Peishen Yan, Shuang Liang, Yang Hua, Linshan Jiang, Kuai Yu, Yulin Sun, Yaozhi Zhang, Tao Song, Ningxin Hu, Xinran Liang, Bingsheng He, Haibing Guan

TL;DR

FWeb3 is presented, a practical Web3-enabled FL framework for incentive-aware training in open environments that supports end-to-end incentive-aware FL with transaction and data-transfer overheads of only 21.3% and 3.4% in WAN.

Abstract

Federated learning (FL) enables collaborative model training over distributed private data. However, sustaining open participation requires incentive mechanisms that compensate contributors for their resources and risks. Enabled by Web3 primitives, especially blockchains, recent FL proposals incorporate incentive mechanisms for open participation, yet most focus primarily on algorithmic design and overlook system-level challenges, including coordination efficiency, secure handling of model updates, and practical usability. We present FWeb3, a practical Web3-enabled FL framework for incentive-aware training in open environments. FWeb3 adopts a modular architecture that separates FL functions from Web3 support services, decoupling the off-chain training and data plane from on-chain settlement while preserving verifiable incentive execution. The framework supports pluggable aggregation and contribution evaluation methods and provides a browser-native DApp interface to lower the participation barrier. We evaluate FWeb3 in real-world settings and show that it supports end-to-end incentive-aware FL with transaction and data-transfer overheads of only 21.3% and 3.4% in WAN; FWeb3 also deploys from zero configuration in under 3 minutes and enables user onboarding in under 1 minute.

FWeb3: A Practical Incentive-Aware Federated Learning Framework

TL;DR

FWeb3 is presented, a practical Web3-enabled FL framework for incentive-aware training in open environments that supports end-to-end incentive-aware FL with transaction and data-transfer overheads of only 21.3% and 3.4% in WAN.

Abstract

Federated learning (FL) enables collaborative model training over distributed private data. However, sustaining open participation requires incentive mechanisms that compensate contributors for their resources and risks. Enabled by Web3 primitives, especially blockchains, recent FL proposals incorporate incentive mechanisms for open participation, yet most focus primarily on algorithmic design and overlook system-level challenges, including coordination efficiency, secure handling of model updates, and practical usability. We present FWeb3, a practical Web3-enabled FL framework for incentive-aware training in open environments. FWeb3 adopts a modular architecture that separates FL functions from Web3 support services, decoupling the off-chain training and data plane from on-chain settlement while preserving verifiable incentive execution. The framework supports pluggable aggregation and contribution evaluation methods and provides a browser-native DApp interface to lower the participation barrier. We evaluate FWeb3 in real-world settings and show that it supports end-to-end incentive-aware FL with transaction and data-transfer overheads of only 21.3% and 3.4% in WAN; FWeb3 also deploys from zero configuration in under 3 minutes and enables user onboarding in under 1 minute.
Paper Structure (43 sections, 1 equation, 4 figures, 6 tables)

This paper contains 43 sections, 1 equation, 4 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Federated Learning
  4. Incentive-aware FL Design Considerations
  5. Transparent and Permissionless Incentive Settlement
  6. Systematic Efficiency under Web3 Constraints
  7. Security of Model Updates for Incentive Integrity
  8. Usability and Extensibility for Real-world Adoption.
  9. Taxonomy of Open-Source FL Platforms and Incentive-Aware FL Prototypes
  10. FWeb3 Design
  11. Overview
  12. FWeb3 Modular Functional Design
  13. Configuration module
  14. Training module
  15. Aggregation module
  16. ...and 28 more sections

Figures (4)

  • Figure 1: End-to-end IPFS transfer throughput for exchanging model updates under CAN and WAN settings.
  • Figure 2: Overview of FWeb3. The framework involves three roles (customer, trainer, and aggregator) and four FL functional modules (configuration, training, aggregation, and contribution evaluation), organized over a Web3 supporting layer with three components (trusted transaction, information communication, and information protection).
  • Figure 3: Demonstration of FWeb3. The user control panel shows the status of all aggregators and trainers, including peer connections, IPFS availability, dataset readiness, and training progress. System logs and key metrics are also displayed for monitoring and debugging purposes.
  • Figure 4: Per-round non-training time overhead under owner-executed (single aggregator) and committee-executed modes (committee size 5/15/25), with 10/20/50/100 trainers. We compare two block generation intervals (5 s vs. 12 s) and two network latency (50 ms vs. 200 ms). Larger committees increase coordination overhead. Here, c. denotes the owner-executed mode, while d.5, d.15, and d.25 denote committee-executed modes with committee sizes of 5, 15, and 25, respectively.