Dynamic Blockchain Sharding for Efficient and Reliable Vector Data Management in Edge Computing

Managing high-dimensional vector data in edge computing environments presents significant challenges due to resource constraints, intermittent connectivity, and security concerns. Traditional vector databases struggle with efficient data distribution and integrity assurance in such decentralized settings. To address these limitations, we propose a dynamic blockchain sharding framework that optimally partitions vector data while maintaining security and performance guarantees. Our approach integrates a lightweight consensus mechanism with adaptive sharding, dynamically adjusting boundaries based on real-time memory utilization and access patterns. By leveraging a scale-free network structure, our system ensures efficient leader selection and reduces coordination overhead, making it well-suited for resource-constrained edge deployments. Preliminary experimental results show that our adaptive sharding technique enhances throughput, processing over one-third of a million vectors per second with minimal latency while maintaining a balanced load distribution.