Skip to content

GPBchain: A GNN-Assisted Framework for Optimizing Sharding Blockchain Reconfiguration

Dec 2026 · IEEE transactions on computers · Vol 75, pp. 4648-4662 · 0 citations · 34 references

TL;DR

An unsupervised sharding blockchain framework based on Graph Neural Network (GNN) called GPBchain is proposed, which achieves explicit joint optimization of CST reduction and inter-shard workload balancing, thereby enhancing the overall throughput of sharding blockchain systems.

Abstract

Blockchain is widely regarded as the foundational infrastructure for Web3. However, its inherent scalability limitations hinder its ability to support large-scale Web3 applications. Sharding technology is one of the most promising approaches to deal with the scalability problem by partitioning accounts into separate shards and allowing each shard to handle a portion of transactions (TXs) in parallel. However, this approach inevitably introduces cross-shard TXs (CSTs), and minimizing CSTs while maintaining inter-shard workload balance remains a challenging trade-off, preventing system throughput from scaling linearly with the number of shards. Existing solutions typically employ heuristic graph partitioning methods during the sharding blockchain reconfiguration phase to reduce CSTs and balance workload across shards. Nevertheless, these methods are often sensitive to parameter settings, exhibit limited generalizability, and struggle to adapt to the rapidly evolving TX patterns. In this paper, we propose an unsupervised sharding blockchain framework based on Graph Neural Network (GNN) called GPBchain, which achieves explicit joint optimization of CST reduction and inter-shard workload balancing. During the reconfiguration phase, GPBchain leverages GNN for account reallocation, taking into account both the ratio of CSTs and inter-shard workload balancing to optimize account distribution. Compared with existing solutions, experimental results conducted on the real Ethereum on-chain data illustrate that GPBchain achieves superior joint optimization of CST reduction and inter-shard workload balancing, thereby enhancing the overall throughput of sharding blockchain systems.

Read PDF

Similar papers

#graph neural networks Open access Dec 2026

Symphony: A Sharded Blockchain for Efficient Smart Contract Execution

Symphony is a sharded blockchain designed for efficient smart contract execution that offloads cross-shard smart contract transactions to Trusted Execution Environments (TEEs), where stateless EVMs perform off-chain execution to bypass multi-round consensus, significantly accelerating processing.

Rong-Kai Zhang, Qiu-Yu Ding, Qian-Yi Liu et al. · 0 citations
Open access Dec 2026

Towards High-Throughput and Low-Latency Heterogeneous Blockchains via RL-Driven Network Sharding

PabShard is proposed, an intelligent sharding framework for heterogeneous blockchain systems that leverages reinforcement learning to adaptively regulate shard operations according to network states and workload conditions and develops an adaptive block generation rate control mechanism that dynamically regulates the p...

Jun-Cheng Ma, Bao-Chao Chen, Chang-Zhi Li et al. · 0 citations
Book Open access Nov 2026

Poros: Secure and Highly Parallel Sharding Blockchain with Optimized Cross-Shard Transaction Processing

Sharding blockchain is a promising approach to improving blockchain scalability by partitioning the network into multiple parallel shards. However, most existing sharding blockchains adopt Byzantine fault tolerance (BFT) as the black-box intra-shard building block, incurring redundant steps. Moreover, a single leader i...

Zhuo-Cheng Pan, Jian-Wei Liu, An-Di Liu et al. · 0 citations
Sep 2026

Analyzing Broker Workload Distribution in Sharding IoT-Blockchain Systems

The convergence of the Internet of Things (IoT) and blockchain technologies enables diverse applications but faces severe scalability bottlenecks that hinder large-scale data interaction. Sharding has emerged as a promising solution by partitioning the blockchain into parallel subnetworks (shards). Among various approa...

Yue Su, Yang Xiang, Kien Nguyen et al. · 0 citations

Related blog posts

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.