Selected work

Representative Papers

SonicDB S6: A Storage-Efficient Verkle Trie for High-Throughput Blockchains

Apr 07, 2026

This work addresses the inefficiency of Ethereum’s Merkle Patricia Trie (MPT) in supporting high-throughput, short-block-interval blockchains due to its large witness size and high storage overhead. To overcome these limitations, the authors design and implement a Rust-based, high-performance Verkle Trie database for the Sonic blockchain. Leveraging Verkle Trie’s fork-free property, they introduce an occupancy-aware node specialization strategy and Delta nodes that record only modified slots, reducing active and archival storage overhead by 98% and 95%, respectively. Further optimizations—including dynamic programming–guided specialization selection, batched updates, multi-threaded commitment computation, and homomorphic Pedersen caching—enable the system to achieve a 2.85× throughput improvement over a persistent Geth Verkle baseline while maintaining a 300-millisecond block interval.

0 citationsRead paper

A Fast Ethereum-Compatible Forkless Database

Dec 04, 2025

Existing Ethereum StateDB implementations exhibit low verification throughput and high storage overhead in fork-free blockchain settings, primarily due to redundant multi-version state management and underlying key-value stores (e.g., LevelDB) ill-suited for linear state evolution. Method: This paper proposes a native state database design tailored for fork-free consensus chains: it eliminates fork-aware multi-versioning in favor of a single-state evolution model and redesigns the storage architecture to enable EVM-compatible, high-performance read/write paths. Contribution/Results: Experimental evaluation demonstrates a 10× improvement in validator node throughput, a 99% reduction in runtime storage footprint, and a 67% decrease in archival node disk usage—significantly outperforming mainstream StateDB implementations. To the best of our knowledge, this is the first native state storage solution specifically engineered for fast-consensus blockchains, achieving unprecedented efficiency, compatibility, and lightweight operation.

0 citationsRead paper

Efficient Forkless Blockchain Databases

Aug 28, 2025

Forkless Layer-1 blockchains suffer from excessive StateDB storage overhead and low transaction throughput due to conventional database designs that retain redundant fork-support mechanisms. Method: This paper introduces the first state database architecture explicitly tailored for forkless consensus. It abandons legacy fork-compatible design patterns and instead integrates state compression, lazy state updates, and a versioned state trie—leveraging the deterministic, fork-free execution environment inherent to forkless chains to optimize data structures. Contribution/Results: Experimental evaluation against the Geth-based Fantom client demonstrates a 100× reduction in storage footprint and a 10× improvement in transaction throughput. Crucially, this work establishes the first systematic decoupling of database design from fork assumptions, thereby proposing a novel paradigm for high-performance, lightweight Layer-1 node implementations.

0 citationsRead paper

Formal Model Guided Conformance Testing for Blockchains

Jan 15, 2025

Blockchain multi-client implementations frequently suffer from forks and security vulnerabilities due to semantic inconsistencies among clients; existing testing approaches fail to detect subtle, divergence-inducing protocol violations. This paper proposes a bimodal collaborative testing framework that tightly integrates formal protocol models (in TLA+ and Promela) into a deterministic blockchain simulation environment, establishing a bidirectional workflow for trace generation and verification—combining symbolic execution, trace comparison, and automated counterexample synthesis. It is the first approach to achieve tight coupling between formal models and simulators, enabling comprehensive detection of protocol semantic deviations. Evaluated on industrial-grade consensus protocols, the framework uncovers multiple critical consistency flaws overlooked by prior testing methods, significantly enhancing semantic consistency assurance across heterogeneous clients.

0 citationsRead paper

Recent publications

Latest Papers

SonicDB S6: A Storage-Efficient Verkle Trie for High-Throughput Blockchains

Apr 07, 2026

This work addresses the inefficiency of Ethereum’s Merkle Patricia Trie (MPT) in supporting high-throughput, short-block-interval blockchains due to its large witness size and high storage overhead. To overcome these limitations, the authors design and implement a Rust-based, high-performance Verkle Trie database for the Sonic blockchain. Leveraging Verkle Trie’s fork-free property, they introduce an occupancy-aware node specialization strategy and Delta nodes that record only modified slots, reducing active and archival storage overhead by 98% and 95%, respectively. Further optimizations—including dynamic programming–guided specialization selection, batched updates, multi-threaded commitment computation, and homomorphic Pedersen caching—enable the system to achieve a 2.85× throughput improvement over a persistent Geth Verkle baseline while maintaining a 300-millisecond block interval.

0 citationsRead paper

A Fast Ethereum-Compatible Forkless Database

Dec 04, 2025

Existing Ethereum StateDB implementations exhibit low verification throughput and high storage overhead in fork-free blockchain settings, primarily due to redundant multi-version state management and underlying key-value stores (e.g., LevelDB) ill-suited for linear state evolution. Method: This paper proposes a native state database design tailored for fork-free consensus chains: it eliminates fork-aware multi-versioning in favor of a single-state evolution model and redesigns the storage architecture to enable EVM-compatible, high-performance read/write paths. Contribution/Results: Experimental evaluation demonstrates a 10× improvement in validator node throughput, a 99% reduction in runtime storage footprint, and a 67% decrease in archival node disk usage—significantly outperforming mainstream StateDB implementations. To the best of our knowledge, this is the first native state storage solution specifically engineered for fast-consensus blockchains, achieving unprecedented efficiency, compatibility, and lightweight operation.

0 citationsRead paper

Efficient Forkless Blockchain Databases

Aug 28, 2025

Forkless Layer-1 blockchains suffer from excessive StateDB storage overhead and low transaction throughput due to conventional database designs that retain redundant fork-support mechanisms. Method: This paper introduces the first state database architecture explicitly tailored for forkless consensus. It abandons legacy fork-compatible design patterns and instead integrates state compression, lazy state updates, and a versioned state trie—leveraging the deterministic, fork-free execution environment inherent to forkless chains to optimize data structures. Contribution/Results: Experimental evaluation against the Geth-based Fantom client demonstrates a 100× reduction in storage footprint and a 10× improvement in transaction throughput. Crucially, this work establishes the first systematic decoupling of database design from fork assumptions, thereby proposing a novel paradigm for high-performance, lightweight Layer-1 node implementations.

0 citationsRead paper

Formal Model Guided Conformance Testing for Blockchains

Jan 15, 2025

Blockchain multi-client implementations frequently suffer from forks and security vulnerabilities due to semantic inconsistencies among clients; existing testing approaches fail to detect subtle, divergence-inducing protocol violations. This paper proposes a bimodal collaborative testing framework that tightly integrates formal protocol models (in TLA+ and Promela) into a deterministic blockchain simulation environment, establishing a bidirectional workflow for trace generation and verification—combining symbolic execution, trace comparison, and automated counterexample synthesis. It is the first approach to achieve tight coupling between formal models and simulators, enabling comprehensive detection of protocol semantic deviations. Evaluated on industrial-grade consensus protocols, the framework uncovers multiple critical consistency flaws overlooked by prior testing methods, significantly enhancing semantic consistency assurance across heterogeneous clients.

0 citationsRead paper