AI-Enhanced Sharding Consensus Algorithm for High-Concurrency Financial Scenarios: An Optimization and Empirical Study
Year: 2025
Type: conference-paper
Abstract: The rapid digital transformation of the financial sector, driven by high-frequency trading and cross-border payments, demands blockchain systems with millisecond-level latency and high throughput. Traditional single-chain and current sharding architectures struggle to meet these requirements due to limitations in efficiency, security, and adaptability. To address this, this paper proposes an AI-enhanced optimization framework for blockchain sharding consensus tailored to financial applications. The framework introduces four core innovations: a dynamic node weighting mechanism to optimize intra-shard consensus and reduce communication overhead; a parallel verification and dynamic routing scheme to accelerate cross-shard transactions; a risk-aware dynamic sharding mechanism to balance security and scalability; and a smart contract–driven self-optimization strategy enabling adaptive adjustment to fluctuating workloads. Empirical evaluation on a customized distributed ledger platform shows substantial improvements-reducing transaction latency from 850 ms to 120 ms, cutting cross-shard delay from 1.2 s to 210 ms, boosting throughput from 3,200 TPS to 12,800 TPS, and lowering the consensus failure rate to 0.03%. These results validate the framework's capability for high-concurrency financial scenarios and highlight its potential for future extensions in cross-chain interoperability, privacy-preserving computation, and predictive resource scheduling.
Authors: Yijie Shi, Wenfeng Li
Institution: Guangdong University of Science and Technology
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7160053371
