ABChain: A Dynamic Sharding Protocol With Balanced Account Partitioning and Adaptive Brokers for IoT Blockchain

Sharding enhances the Internet of Things (IoT) blockchain scalability by parallel transaction (TX) processing, yet cross-shard TXs and dynamic workload imbalance persist as key bottlenecks. Existing solutions employ brokers that have segmented accounts across multiple shards to efficiently handle cross-shard TXs. Nevertheless, they rely on preset optimal accounts as brokers, which may not be feasible in dynamic IoT blockchains where TX workloads are unpredictable. To this end, this paper proposes ABChain, a blockchain sharding protocol for efficient state synchronization with two key components. One is a balanced account partitioning method which allocates accounts to the shards with the highest degree of correlation, and the other is an adaptive broker framework which includes dynamic broker selection from historically active accounts, seamless broker updates via account splitting/aggregation, and dual broker queues to minimize cross-shard TX overhead. The theoretical analysis shows that the former can reduce the number of cross-shard TXs and mitigate workload imbalance, while the latter can efficiently handle cross-shard TXs and adapt to the dynamic blockchain environments. Experimental evaluations demonstrate ABChain’s superiority. Compared to state-of-the-art protocols, it improves throughput by 33.4%, reduces cross-shard TX rate by 52.3%, and lowers confirmation latency by 67.7%.