ESUM: an efficient UTXO schedule model

The current size of UTXO set in Bitcoin is large and continues to grow, resulting in high memory usage. This paper proposes an efficient UTXO schedule model to lower memory usage based on the observation of UTXO lifespan. Full nodes with the UTXO schedule model store the full UTXO set in disk database and schedule a subset of full UTXO set in memory, converting UTXO querying and accessing into multi-layer activities. Machine learning methods are applied to train a well-performed model based on historical transaction data and market trade data. Experiment results indicate that the UTXO schedule model could reduce nearly 90% memory usage compared to original Bitcoin nodes, allowing more nodes with different memory capability to participate in the blockchain.