Joint Optimization of Node Selection and Resource Allocation for Sharded Blockchain

To enhance the scalability of sharded blockchain, we propose a joint optimization scheme for node selection and resource allocation. For transactions with varying computational demands, we use queuing theory to analyze the computation and sojourn time of transaction verifications. Based on this analysis, we construct an optimization model to maximize throughput and minimize latency, where the discrete node allocation and continuous resource allocation are separated and iterated using the Benders decomposition technique. These are further solved using the successive approximation approach and the branch-and-bound algorithm, respectively. Simulations show that the proposed scheme can significantly improve the throughput and reduce the latency of the blockchain system.