Layer-1 Architecture
Research Papers
Academic and industry research
Sei: The Layer 1 for Trading
Jayendra Jog, Jeff Feng, Dan Edlebeck
Sei is presented as a sector-specific Layer 1 blockchain optimized specifically for the needs of decentralized finance (DeFi) and high-frequency trading applications. Recognizing that trading platforms require low latency and high reliability, the Sei whitepaper introduces an integrated order-matching engine and a parallelized processing architecture. By optimizing the consensus protocol for frequent state updates, Sei minimizes the time to finality, a critical requirement for market-making and order-book stability. The architecture includes a native matching engine that replaces the need for external smart contract-based order books, thereby reducing gas overhead and slippage. The paper highlights the protocol's ability to handle high order volume through a combination of fast consensus and optimized transaction ordering. This document is significant for its application-specific approach to blockchain design, suggesting that general-purpose L1s may face limitations in highly competitive trading environments.
PHANTOM and GHOSTDAG: A Scalable Generalization of Nakamoto Consensus
Yonatan Sompolinsky, Shai Wyborski, Aviv Zohar
This paper introduces PHANTOM and GHOSTDAG, two protocols that generalize the Nakamoto consensus to directed acyclic graph (DAG) topologies. The traditional blockchain structure restricts throughput by forcing a linear order of blocks, often leading to stale blocks and network inefficiency. The authors present a mechanism where blocks can be mined in parallel and integrated into a DAG, while still ensuring a robust total ordering of transactions. PHANTOM provides a theoretical solution for a BFT-like consensus in a PoW-based network, while GHOSTDAG offers a scalable, highly efficient implementation that allows for fast confirmation times without compromising security. These contributions are instrumental in understanding how to maintain the decentralization properties of Proof of Work while breaking the scalability constraints of linear chains. The analysis provides deep insights into block-DAG structures and the mathematical proofs required to guarantee network security in non-linear topologies.
Prism: Deconstructing the Blockchain to Approach Physical Limits
Vivek Bagaria, Sreeram Kannan, David Tse, Giulia Fanti, Pramod Viswanath
Prism is a theoretical and practical proposal that deconstructs the traditional blockchain structure into modular components to reach the maximum limits of throughput and latency governed by physical network properties. The authors propose a parallelized block structure where miners produce blocks for different aspects of the protocol independently, rather than as a monolithic chain. By separating the roles of leader election, transaction processing, and data availability, the system approaches the performance limits imposed by bandwidth and propagation delays. The paper rigorously demonstrates how this decomposition allows the network to handle an order of magnitude higher throughput than current consensus protocols while maintaining strong security guarantees. The model essentially scales with the available bandwidth of the nodes, rather than being restricted by the slowest participant or the consensus round time. Prism provides a critical blueprint for designing future protocols that prioritize performance as a fundamental constraint.