RT-DAG: DAG-Based Blockchain Supporting Real-Time Transactions

The nonlinear structure of a directed acyclic graph (DAG)-based blockchain facilitates parallel transaction processing, improving the overall scalability and throughput of the system. Compared with chain-based structures, DAG-based blockchains offer more flexibility in transaction processing. However, current research predominantly concentrates on improving throughput and reducing latency, with insufficient attention given to the real-time characteristics of transactions. Consequently, these systems struggle to effectively support business scenarios with deadlines, such as supply chain management and engineering manufacturing. In this article, we present a prototype of a DAG-based blockchain called RT-DAG, which supports the execution of real-time transactions. Specifically, RT-DAG assigns priorities to transactions based on deadlines and captures address utilization during transaction execution via an address resource graph (ARG). To enhance address utilization efficiency and provide additional opportunities to commit transactions before deadlines, we introduce a concurrency control scheme based on ARG called opportunity seizing (OS), which involves reordering transactions to replenish available addresses. Additionally, we propose a load-balancing grouping strategy to alleviate single-node workload pressures. The strategy assesses each group's workload based on transaction quantity and type. Leveraging the bat algorithm, the algorithm strives to balance the workload distribution among groups, ultimately improving transaction execution efficiency. Numerous experiments have shown that RT-DAG performs well under different conflict rates and deadline ranges, effectively meeting the transaction timeliness requirements.