CAVer: Compacting accumulation to reduce cross-chain verification overhead

The data silos between heterogeneous blockchain platforms have given rise to cross-chain protocols. These cross-chain protocols enable secure interoperability between multiple distinct blockchains. Security requires that the target chain has the ability to verify whether cross-chain messages have been finalized on the source chain. Existing cross-chain protocols first synchronize source chain block headers to the target chain and validate them, then verify cross-chain messages included in the header's commitment value using native accumulators, like a Merkle tree. Cross-chain verification incurs significant on-chain overhead. Existing solutions primarily focus on optimizing header verification, while neglecting verification overhead for messages. It is worth noting that as the number of on-chain states increases, the number of elements in the accumulator multiplies, leading to a larger proof size that escalates message verification overhead on the target chain. To reduce message verification overhead on the target chain, we present CAVer. It removes non-cross-chain states from accumulators to minimize message proof size. CAVer incorporates three schemes: Fill-Weak, Fill-Mid, and Fill-Pro. Fill-Weak abandons blockchain's native accumulators by creating a new contract-based compacting accumulator exclusively for cross-chain messages on the source chain. Based on Fill-Weak, Fill-Mid, and Fill-Pro employ block-isolation and target-isolation, respectively, to further reduce verification overhead. Additionally, we design the algorithm AppendAndTailor to reduce storage cost on the source chain. It aims at trimming historical states when accumulating cross-chain messages. Evaluations on Ethereum-like blockchains demonstrate that CAVer reduces message verification overhead on the target chain by 76% compared to the existing state-of-the-art scheme.