Downsampling and Transparent Coding for Blockchain

Abstract

Blockchain is considered to be able to solve the problem of trust between distributed nodes. However, it's a challenge to store the large amount of data produced by blockchain. Therefore, it is necessary to reduce node storage overhead while ensuring node independence and data recoverability. This paper proposes to downsample these data to reduce the storage overhead of nodes. These nodes keep good independence, if downsampling follows the entropy of blockchain and the block bodies are continuously stored. Moreover, it demonstrates that the entire blockchain history can be efficiently recovered through the cooperative decoding of a group of nodes like fountain codes, if reserved data over these nodes obey the soliton distribution. However, these data on nodes are transparent (stored data are uncoded and can be used directly by the node without decoding). Thus, the proposed algorithm not only keeps decentralization and security, but also has good scalability in independence and recoverability.