[Paper Review] Decentralized Federated Learning: Fundamentals, State of the Art, Frameworks, Trends, and Challenges

Enrique Tomás Martínez Beltrán, Mario Quiles Pérez+6 more|arXiv (Cornell University)|Nov 15, 2022

Privacy-Preserving Technologies in Data206 references57 citations

TL;DR

The paper surveys Decentralized Federated Learning (DFL), clarifying fundamentals, taxonomy, frameworks, application scenarios, and emerging trends, lessons, and open challenges. It differentiates DFL from CFL and maps frameworks to use cases.

ABSTRACT

In recent years, Federated Learning (FL) has gained relevance in training collaborative models without sharing sensitive data. Since its birth, Centralized FL (CFL) has been the most common approach in the literature, where a central entity creates a global model. However, a centralized approach leads to increased latency due to bottlenecks, heightened vulnerability to system failures, and trustworthiness concerns affecting the entity responsible for the global model creation. Decentralized Federated Learning (DFL) emerged to address these concerns by promoting decentralized model aggregation and minimizing reliance on centralized architectures. However, despite the work done in DFL, the literature has not (i) studied the main aspects differentiating DFL and CFL; (ii) analyzed DFL frameworks to create and evaluate new solutions; and (iii) reviewed application scenarios using DFL. Thus, this article identifies and analyzes the main fundamentals of DFL in terms of federation architectures, topologies, communication mechanisms, security approaches, and key performance indicators. Additionally, the paper at hand explores existing mechanisms to optimize critical DFL fundamentals. Then, the most relevant features of the current DFL frameworks are reviewed and compared. After that, it analyzes the most used DFL application scenarios, identifying solutions based on the fundamentals and frameworks previously defined. Finally, the evolution of existing DFL solutions is studied to provide a list of trends, lessons learned, and open challenges.

Motivation & Objective

• Identify and define the fundamental elements that differentiate DFL from CFL (architecture, topology, communication, security, KPIs).
• Catalog and compare existing open-source frameworks enabling DFL.
• Analyze DFL application scenarios (healthcare, Industry 4.0, mobile, military, vehicles) and their requirements.
• Extract trends, lessons learned, and open challenges to guide future research and practice.

Proposed method

• Review and synthesize literature on DFL fundamental components (federation architectures, topologies, communication mechanisms, security/privacy, KPIs).
• Develop a taxonomy and framework to categorize DFL solutions (based on architecture, topology, data distribution, roles, decentralization).
• Survey open-source DFL frameworks and map them to use-case applicability.

Experimental results

Research questions

• RQ1 What are the fundamental aspects of DFL (architecture, topology, communication, security, KPIs) and how are they combined in solutions?
• RQ2 Which DFL frameworks exist and what fundamentals do they provide for building DFL solutions?
• RQ3 What are the main characteristics of the most relevant DFL application scenarios?
• RQ4 What trends, lessons learned, and challenges have emerged in DFL?

Key findings

• DFL fundamentals encompass federation architecture, network topology, communication mechanisms, security/privacy, KPIs, and optimization techniques.
• DFL architectures are categorized by federation type (cross-silo vs cross-device), participant roles (trainer, aggregator, proxy, idle), and decentralization schema (DFL, SDFL, CFL).
• A taxonomy and comparative table summarize how different works address data distribution, topology, communication, and security aspects in DFL.
• The paper lists representative application scenarios (healthcare, Industry 4.0, mobile services, military, vehicles) and discusses how solutions address them.
• It provides a synthesis of trends, lessons learned, and open challenges to guide future DFL research and practice.
• The work positions itself as a comprehensive literature review and taxonomy for DFL, filling gaps left by prior surveys that focused on CFL or narrower aspects.