Reducing Storage Requirement in Blockchain via Node-Oriented Block Placement

Blockchain nodes need to store all the transactions from the beginning, which places heavy storage loads and hinders the scalability of blockchain systems. It is difficult to directly apply blockchain to real environments where nodes have limited resources. Prior work alleviated the storage challenge by letting engaged nodes hold a portion of the complete blockchain data. However, most existing solutions placed blockchain data onto nodes based on data content. Those studies neglected node attributes such as storage capacity and query cost, thereby failing to utilize nodes' storage space and computational power effectively. To address this issue, we propose NOBS, a node-oriented block storage scheme that reduces storage requirements in blockchain by optimizing the placement of blocks based on node attributes. We introduce the concept of assignment gain using a weighted function on node attributes. We formally define the block assignment problem that maximizes the overall assignment gain under the storage limit of each node. We present two heuristic algorithms to solve this NP-hard problem. We also address the block assignment problem from the perspective of multi-objective optimization and propose an algorithm based on skyline layers. Extensive experimental evaluation shows the promising results of our proposed approaches.