LiteChain: A Lightweight Blockchain for Verifiable and Scalable Federated Learning in Massive Edge Networks

TL;DR

LiteChain presents a lightweight blockchain framework for verifiable and scalable federated learning over Massive Edge Networks by introducing a two-tier clustering that enables efficient intra- and inter-cluster training. It combines off-chain verification within clusters with on-chain Comprehensive Byzantine Fault Tolerance consensus for inter-cluster updates, plus an update-consensus mechanism to prune stale data and refresh committees. Theoretical analysis demonstrates reduced communication complexity and convergence guarantees under asynchronous hierarchical FL, and extensive experiments show LiteChain achieves lower latency and storage overhead while resisting replay and data-poisoning attacks. The approach offers practical scalability for secure edge intelligence, with future work targeting enhanced privacy protections such as differential privacy or homomorphic techniques.

Abstract

Leveraging blockchain in Federated Learning (FL) emerges as a new paradigm for secure collaborative learning on Massive Edge Networks (MENs). As the scale of MENs increases, it becomes more difficult to implement and manage a blockchain among edge devices due to complex communication topologies, heterogeneous computation capabilities, and limited storage capacities. Moreover, the lack of a standard metric for blockchain security becomes a significant issue. To address these challenges, we propose a lightweight blockchain for verifiable and scalable FL, namely LiteChain, to provide efficient and secure services in MENs. Specifically, we develop a distributed clustering algorithm to reorganize MENs into a two-level structure to improve communication and computing efficiency under security requirements. Moreover, we introduce a Comprehensive Byzantine Fault Tolerance (CBFT) consensus mechanism and a secure update mechanism to ensure the security of model transactions through LiteChain. Our experiments based on Hyperledger Fabric demonstrate that LiteChain presents the lowest end-to-end latency and on-chain storage overheads across various network scales, outperforming the other two benchmarks. In addition, LiteChain exhibits a high level of robustness against replay and data poisoning attacks.

Figures (12)

• Figure 1: The system overview of LiteChain, including the architecture of LiteChain (left) and LiteChain application in a sample MEN (right). The architecture includes: ➀ Task publishers dispatch FL tasks to edge devices; ➁ A two-tier LiteChain formation via distributed optimization algorithm; ➂ Off-chain verification-based intra-cluster local training; ➃ Request on-chain verification via consensus mechanism; ➄ Block invocation; ➅ LiteChain synchronization; ➆ Aggregation of updated models according to LiteChain; ➇ After $n$ step updates leading to reputation record updates and redundant storage clearance via commitment.
• Figure 2: The workflow of LiteChain.
• Figure 3: Accuracy over time (in seconds) with IID dataset.
• Figure 4: Accuracy over time (seconds) with non-IID dataset.
• Figure 5: Latency overhead of two specific tasks: Training Task (TT) and Verification Task (VT) in one-round FL training across 50-300 devices.

Theorems & Definitions (12)

• Theorem 1
• Theorem 2
• Theorem 3
• Lemma 1
• Lemma 2
• Theorem 4
• proof
• proof
• proof
• proof