Adventures of a Light Blockchain Protocol in a Forest of Transactions: A Subset of a Story

The Ethereum blockchain is one of the most popular permissionless blockchains. A consequence of its popularity has been the growth of processing and data storage requirements for any node participating in the Ethereum blockchain network. For constrained devices such requirements are often infeasible to meet. To cater for such nodes, a so-called light protocol has been proposed for Ethereum where the responsibility of maintaining a correct state representation is delegated to light protocol servers. Previous research has identified dependence on external state management as a potential security vulnerability that exclusively impacts light nodes. Although a simple mitigation strategy is available, it comes at the expense of increased latency. In this work, we propose a new Ethereum node type, which we call a subset node, as an extension of the light protocol. Our proposal allows subset nodes to gain a lower latency than a pure light node with comparable or even higher security assurances by tracking and evaluating only a subset of all of the transactions issued on the blockchain. We provide a formal proof on the correctness of the blockchain state used by the subset node under the proposed model. To evaluate the practical feasibility of the subset node model, we analyze one year of historical transaction data from Ethereum, and demonstrate that a subset node tracking the state of a single account can achieve a significant reduction in storage and computational requirements when compared to a full node.