Downsampling Blockchain Algorithm

Abstract

In blockchain, every full node has to store all history transactions in block bodies. However, with the rapid growth of transactions, the storage bloating problem has emerged, and made it challenging for a mobile system to afford the storage and synchronization overhead. This paper proposes to downsample block bodies to reduce the nodes' storage. By downsampling block bodies by M times, only about 1/M of the block bodies with high information entropy are downloaded and stored. Performance analysis demonstrates that the proposed downsampling nodes and full nodes have similar broadcast accuracy with appropriate M. The simulation results show that the proposed algorithm can provide a better cost-effective choice for nodes between broadcast accuracy and storage.