An Investigation of Broker Selectionfs Impact on Sharding IoT-Blockchain Performance <Abstract> 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 approaches, Broker-based sharding (e.g., BrokerChain) enhances scalability by utilizing Broker accounts to convert complex cross-shard transactions into intra-shard ones. However, existing research typically relies on fixed, preset accounts as Brokers, lacking a rigorous analysis of the rationale behind this selection and its impact on system performance. Furthermore, alternative Broker account selection strategies remain largely unexplored. To address this gap, this study analyzes account characteristics within historical datasets to determine the origin of effective Broker accounts. We design and evaluate three distinct selection strategies: high-activity-based, low-activity-based, and hybrid-activity-based. These strategies are implemented and tested on a Broker-based sharding IoT-Blockchain system using the BlockEmulator platform. We assess their impact on key performance metrics, including Transactions Per Second (TPS), Confirmation Latency, and Cross-shard Transaction Ratio (CTX), as well as Broker load distribution. Experimental results demonstrate that the high-activity-based selection achieves the superior performance, despite exhibiting a load imbalance, which remains more favorable than the alternatives. Conversely, the low-activity-based method yields the lowest throughput, while the hybrid approach offers moderate performance but suffers from noticeable load imbalance. |