Blockchain Sharding Algorithm Based on Account Degree and Frequency

The long transaction latency and low throughput of blockchain are the key challenges affecting the large-scale adoption of blockchain technology. Sharding technology is a primary solution by divides the blockchain network into multiple independent shards for parallel transaction processing. However, most existing random or modular schemes fail to consider the transactional relationships between accounts, which leads to a high proportion of cross-shard transactions, thereby increasing the communication overhead and transaction confirmation latency between shards. To solve this problem, this paper proposes a blockchain sharding algorithm based on account degree and frequency (DFSA). The algorithm takes into account both account degree and weight relationships between accounts. The blockchain transaction network is modeled as an undirected weighted graph, and community detection algorithms are employed to analyze the correlations between accounts. Strong-correlated accounts are grouped into the same shard, and a multi-shard blockchain network is constructed. Additionally, to further reduce the number of cross-shard transactions, this paper designs a random redundancy strategy based on account correlation, which randomly selects strong-correlated accounts and stores them redundantly in another shard, thus original cross-shard transactions can be verified and confirmed within the same shard. Simulation experiments demonstrate that DFSA outperforms the random sharding algorithm (RSA), modular sharding algorithm (MSA), and label propagation algorithm (LPA) in terms of cross-shard transaction proportion, latency, and throughput. Therefore, DFSA can effectively reduce cross-shard transaction proportion and lower transaction confirmation latency.