一种新的、“更好的”加密货币Unit-e，能在一秒内执行五千至一万笔交易（Visa级），同时还能保证链上和网络安全性，这是MIT、Stanford、UC Berkeley、USC、UIUC、CMU、UW Seattle七家高校的学者在基金会DTR的资助下提出的。

　　根据他们的报告，他们设计了新的共识算法、P2P网络算法、分片算法、路由算法，重点放在了扩展性（scalability），实现了安全性、存储、计算方面的最佳权衡，达到了现有物理限制下的极限。

　　这是学界对币圈的最新动议，是争取去中心化的新一次尝试。

　　不多说，以下提供官方新闻稿和报告《去中心化支付系统：原理及设计》第一、二章概述部分（pp. 8-18）的翻译，中英对照。

　　分布式技术研究成立基金会并推出电子单元（Unit-e），一个全球扩展的分散支付网络 Distributed Technologies Research Launches Foundation and Introduces Unit-e， a Globally Scalable Decentralized Payments Network

　　顶尖学者联手应对未解决的区块链扩展性的挑战 Leading Academics Join Forces to Tackle Unsolved Challenges of Blockchain Scalability

　　瑞士ZUG，2019年1月17日，PRNewswire讯 分布式技术研究（Distributed Technologies Research，DTR）是一家为分布式技术的创新研究和开发提供资金的非营利基金会，今天正式宣布启动，并推出其首个项目Unit-e，一个全球扩展的去中心化支付系统。

　　ZUG， Switzerland， Jan. 17, 2019 /PRNewswire/ -- Distributed Technologies Research（"the Foundation" or "DTR"）， a non-profit foundation that funds innovative research and development of distributed technologies， today announced its official launch， and introduces its first project， Unit-e， a globally scalable decentralized payments system.

　　因扩展性的挑战，加密货币尚未得到广泛采用。DTR正资助研究，处理区块链技术全部堆栈，以提供可扩展功能。这些创新将被用于正在开发的Unit-e，并适用于所有通用型区块链。

　　Cryptocurrencies have not yet achieved widespread adoption， in part due to scalability challenges. DTR is funding research to address the entire blockchain technology stack to deliver scalable performance. These innovations are being applied in the development of Unit-e and are applicable to all general-purpose blockchains.

　　“区块链和数字货币市场处于一个有趣的十字路口，让人回想起电信和互联网等行业即将达到拐点的时刻，”DTR基金会理事会主席Babak Dastmaltschi说，“这都是变革性的时代。将世界上每个人联系在一起，我们已十分接近。分布式技术的进步将使开放式网络成为可能，无需中心权威的管理。DTR的成立旨在实现和支持这场革命，正是在这种情况下，我们推出了Unit-e。“

　　"The blockchain and digital currency markets are at an interesting crossroads， reminiscent of the inflection points reached when industries such as telecom and the internet were coming of age，" said Babak Dastmaltschi， Chairman of the DTR Foundation Council. "These are transformative times. We are nearing the point where every person in the world is connected together. Advancements in distributed technologies will enable open networks， avoiding the need for centralized authorities. DTR was formed with the goal of enabling and supporting this revolution， and it is in this vein that we unveil Unit-e."

　　分布式技术研究 Distributed Technologies Research

　　是这个信念让我们成立该基金会：分布式信任将成为我们社会的核心，并且需要创新的新研发来实现这一目标。它的使命是：将网络、分布式存储、信息理论、通信理论、激励设计、博弈论和密码学等不同领域的全球思想领袖、开发人员、研究人员联系在一起，促进业界与学术界的大规模协作。

　　The Foundation was established with the belief that distributed trust will be at the core of our society and that innovative new research and development is needed to achieve this. Its mission is to promote large-scale collaboration across industry and academia by linking together a global community of thought leaders， developers and researchers across fields ranging from networking， distributed storage， information theory， communication theory， incentive design， game theory and cryptography.

　　DTR正在资助来自美国顶尖大学的研究人员的分布式协作，致力于解决区块链可扩展性未解决的技术挑战。该研究小组将终身教授与冉冉升起的学术新星、获奖者、专利持有着、作者和成功的分布式技术企业家融合一起。他们每一位都将个体的专业知识带到不断发展的跨学科团队，专注于使用第一性原则、全栈方法从头开始重新设计区块链。

　　DTR isfunding a distributed collaboration of researchers from top universities acrossthe US， working on the unsolved technical challenges of blockchain scalability. The research group mixes tenured professors with rising research stars， award winners， patent holders， authors and successful distributed technologies entrepreneurs. They each bring their individual subject matter expertise to the growing， multi-disciplinary team focused on redesigning blockchains from the ground up with a first-principles， full-stack approach.

　　“在比特币首次出现后的10年里，区块链已经从一种新颖的想法发展成为一个学术研究领域，”DTR首席研究员、卡内基梅隆大学电子与计算机工程助理教授Giulia Fanti说，“我们的方法是首先理解区块链性能的基本限制，然后以在严格理论框架内可证为要求，发现解决方案，尽可能接近这些限制。”

　　"In the 10 years since Bitcoin first emerged， blockchains have developed from a novel idea to a field of academic research，" said Giulia Fanti， a leadresearcher for DTR and Assistant Professor of Electrical and Computer Engineering at Carnegie Mellon University. "Our approach is to first understand fundamental limits on blockchain performance， then to develop solutions that operate as close to these limits as possible， with results that are provable within a rigorous theoretical framework."

　　作为非营利组织，所有基金会的资源必须用于支持研究和开发。DTR致力于，在全球范围内，持续支持和资助研究合作和Unit-e开源开发者社区。

　　As a non-profit， all the Foundation's resources must be directed towards supporting research and development. DTR is committed to growing both the research collaboration and the Unit-e open-source developer community globally withongoing support and funding.

　　“DTR正在创造一个良性循环的生态系统，”DTR的研究员、伊利诺伊大学厄巴纳-香槟分校电子与计算机工程教授Pramod Viswanath说，“创新的研究吸引了顶尖学者参与合作，这进一步加快了创新步伐。迄今为止，我们已有超过10篇论文在同行评审的顶级学术组织发表（或正在审核中）。”

　　"DTR is creating an ecosystem with a virtuous cycle，" said Pramod Viswanath， a researcher for DTR and Professor of Electrical and Computer Engineering at the University of Illinois Urbana-Champaign. "Innovative research attracts top caliber academics to the collaboration， which further accelerates the pace ofinnovation. This has resulted in over 10 papers so far， published （or underreview） in peer-reviewed top scientific venues."

　　迄今为止发布的解决方案旨在形成一个整体方案，解决区块链技术堆栈的每一层：共识、分片、支付渠道、隐私与安全，及经济学与激励。DTR关于区块链完善研究计划的愿景载于研究宣言《去中心化支付系统：原理及设计》，其中介绍了Unit-e的架构，并将作为一部研究著作出版。

　　The solutions published to date seek to form a holistic package addressing each layer of the blockchain technology stack: consensus， sharding， payment channels， privacy and security， and economics and incentives. DTR's vision for a consummate research program on blockchains is set out in a research manifesto， "Decentralized Payment Systems: Principles and Design，" which introduces the architecture of Unit-e and will be published as a research book.

　　电子单元 Unit-e

　　货币是世界上最为通行的制度性信任体系，但对这些金融机构的信任已被侵蚀。需要一种新方法，本基金会认为支付是区块链首个“杀手级应用”。因此，DTR的首先推出的是全球扩展的去中心化支付网络：Unit-e。

　　Money is the world's most universal system of institutional trust， but trust in these financial institutions has been eroded. A new approach is needed， and the Foundation believes payments is the first "killer application" forblockchains. Therefore， DTR's first initiative is a globally scalable decentralized payments network: Unit-e.

　　“扩展性的缺乏阻碍了加密货币的应用，而DTR的开创性研究正在解决这个问题，”加入DTR基金会理事会的Joey Krug说道，他是Unit-e的支持者Pantera Capital的联合首席投资官，“Unit-e开发人员正在将这项研究转化为真正的可扩展性能，这将有助大量分散的金融应用。”

　　"A lack of scalability is holding back cryptocurrency adoption， and DTR's groundbreaking research is addressing this，" said Joey Krug， who is joining the DTR Foundation Council， and is Co-Chief Investment Officer at Pantera Capital， a backer of Unit-e. "The Unit-e developers are turning this research into real scalable performance which will benefit a huge swath of decentralized financial applications."

　　Unit-e核心是位于德国柏林的开源和分布式系统工程师顶级团队。团队成员来自8个国家，在多行业有构建协议、开发人员工具、API和高扩展软件10年的平均经验。该团队确保Unit-e代码库保持创新、代码质量、安全性、协议设计和开源准备度的最高标准。

　　Unit-e core is a top-tier team of open-source and distributed systems engineers based inBerlin， Germany. The team represents 8 nationalities with an average of 10 years of experience building protocols， developer tools， APIs and high-scale software across multiple industries. The team ensures the Unit-e codebase maintains the highest standards for innovation， code quality， security， protocol design and open-source readiness.

　　该项目的意识形态牢牢植根于透明度，信赖开源，以包容性决策为了公共利益而开发去中心化软件。善治对于维护此意识形态和促进Unit-e生态系统的长期发展至关重要。

　　The project's ideology is firmly rooted in transparency， with a belief in open-source， decentralized software developed in the public interest with inclusive decision-making. Good governance is essential for maintaining this ideology and fostering the long-term development of the Unit-e ecosystem.

　　“拥有一个健康有效的治理流程对于加密货币项目的成功非常重要，”Unit-e独立技术指导委员会主任、伊利诺伊大学香槟分校电子与计算机工程系助理教授Andrew Miller说，“Unit-e从一开始就专注于解决这些问题，从成功的开源和学术项目的悠久历史中吸取教训。”

　　"Having a healthy and effective governance process is important for the success of a cryptocurrency project，" said Andrew Miller， head of the Unit-e independent technical steering committee and Assistant Professor of Electrical and Computer Engineering at University of Illinois Urbana-Champaign." Unit-e has focused on addressing these topics from the start， learning lessons from a long history of successful open-source and academic projects."

　　Unit-e的目标是在2019年下半年全网启动。

　　Unit-e is targeting a network launch in the second half of 2019.

　　关于分布式技术研究 About Distributed Technologies Research

　　分布式技术研究（“基金会”或“DTR”）是一家总部位于瑞士的非营利性基金会，为分布式技术的创新研究和开发提供资金。其首个项目是加密货币Unit-e。由突破性研究支持，Unit-e是一个全球扩展的去中心化支付网络。欲了解更多信息，请访问www.dtr.org。

　　Distributed Technologies Research （"the Foundation" or "DTR"） is a Swiss-based non-profit foundation that funds innovative research and development of distributed technologies. Its first initiative is the cryptocurrency Unit-e. Backed by ground breaking research， Unit-e is a globally scalable decentralized payments network. For more information， please visit www.dtr.org.

　　消息来源分布式技术研究 SOURCE Distributed Technologies Research

　　相关链接 http://www.dtr.org

　　去中心化支付系统：原则与设计

　　Decentralized Payment Systems: Principles and Design

　　Editors: Giulia Fanti and Pramod Viswanath

　　The Distributed Technology Research Foundation

　　January 16,2019

　　第一章 导论 Introduction

　　Giulia Fanti，CMU

　　Pramod Viswanath，UIUC

　　Unit-e是一种专门用于支付的加密货币，强调性能和最先进的去中心扩展性。虽然提供智能合约的区块链必须解决完整、去中心的状态机复制问题，但轻量级支付交易提供了大量的并行性。因此，我们缩小了通用型区块链系统的目标，并以整体的第一性原则方式处理这些问题。我们注意到，虽然重点关注的是支付，但我们发现Unit-e设计中涉及的大部分研究成果也提供了可编程性（例如通过用图灵或伪图灵完整语言编写的智能合约）；我们将在宣言的适当部分讨论这些后果。为了成为普在的全球支付系统，Unit-e旨在以完全去中心的方式满足以下五个要求：

　　Unit-e is a cryptocurrency that specializes exclusively on payments， with a strong emphasison performance and state-of-the-art， decentralized scalability. Whereas blockchains offering smart contracts must solve the problem of full， decentralized state-machine replication， lightweight payment transactions offer massive parallelism. Consequently， we narrow the objectives of a general purpose blockchain system， and tackle these questions in a holistic， first-principles manner. We note that although focused on payments， we find that much of the research output involved in the design of Unit-e also provides programmability （e.g. through smart contracts written in Turing or pseudoTuringcomplete languages）； we discuss these ramifications in the appropriate sections throughput the manifesto. To become a ubiquitous global payment system， Unit-e is designed to meet the following five requirements in a fully-decentralized manner:

　　1.安全。 系统应拒绝执行未经授权或无效的支付。

　　1.Security. The system should prevent unauthorized or invalid payments from being executed.

　　2.延迟。 交易应在秒的时间范围内无缝处理。

　　2.Latency. Transactions should be processed seamlessly， on the timescale of seconds.

　　3.吞吐量。 整个网络应该能够每秒确认多达数千个交易。

　　3.Throughput. The network as a wholeshould be able to confirm up to thousands of transactions per second.

　　4.可用性。 该系统应始终可访问，提供低廉且可预测的交易费用及运营网络的低成本，并提供无缝且可预测的用户体验。

　　4.Usability. The system should be accessible at all times， offer low and predictable fees and a low cost ofoperating the network and provide a seamless and predictable user experience.

　　5.隐私。 系统应防止未授权方访问交易日志。

　　5.Privacy. The system should prevent unauthorized parties from accessing transaction logs.

　　关键的挑战是在高效，可扩展，去中心化的平台中满足这些要求。此外，这五个基本属性之间的权衡通常是质化的，很少有工作明确地建模或分析它们之间的基本权衡。为了弥合这一差距，基金会正在促进更广泛的跨学术界、非营利部门、企业界和研究团体的研究，使他们参与Unit-e的发展。

　　A key challenge is to meet these requirements in an efficient， scalable， decentralized platform. Furthermore， the trade-offs between these five essential properties are often presented qualitatively， with little work explicitly modeling or analyzing the fundamental trade-offs among them. In an effort to bridge this gap， the Foundation is catalyzing research across the broader academic， non-profit， enterprise and research communities， engaging them in the development of Unit-e.

　　该宣言代表了此类研究活动的输出，并突出了对构建真正可扩展的去中心支付系统至关重要的关键见解。具体而言，我们从信息理论、网络、编码理论、博弈论和经济学中汲取灵感，此外还有与分布式系统和密码学更为令人熟知的联系。

　　This manifesto represents the output of such a research engagement and it highlights key insights that are fundamental to the building of a truly scalable decentralized payment system. Specifically， we draw inspiration from information theory， networking， coding theory， game theory， and economics， in addition to more well-known connections to distributed systems and cryptography.

　　1.1要求 Requirements

　　我们将更详细地重新审视我们的要求，并提供基准性能数据，开始本章。然后，我们将讨论提供所欲性能指数所需的技术构建模块，制定与区块链相关的核心科学和工程的基础和根本问题。

　　We begin this chapter by revisiting our requirements in greater detail and providing benchmark performance figures. Then， we discuss the technical building blocksneeded to deliver the desired figures of merit， formulating basic and fundamental questions of core scientific and engineering relevance to blockchains.

　　安全性。如果轻易实施，区块链系统能引入许多与可用性、实施问题、激励和容错相关的安全性问题。从算法设计的角度来看，我们希望防范两种主要类型的安全违规。第一种是未经授权的用户使用其他用户的资金进行支付。第二个涉及用户双重支付自己的钱。这两种攻击都是盗窃，但第二种威胁是数字货币独有的。给定一安全的代码库，非对称密码术的使用自然地防止了未授权交易。另一方面，双重花费应以去中心的共识机制予以特别防止。在低延迟、高吞吐量系统中应对这一挑战，尤具挑战性，这也是Unit-e设计的重点。

　　Security. If implemented naively， blockchain systems can introduce many security concerns related to usability， implementation issues， incentives， and fault-tolerance. From an algorithmic design standpoint， we wish to protect against two main types of security violations. The first involves unauthorized users making payments with other users’ funds. The second involves users double spending their own money. Both attacks amount to theft， but the second threat is unique to digital money. Assuming a secure codebase， unauthorized transactions are organically prevented through the use of asymmetric cryptography. Double spends， on the other hand， should be specifically prevented by decentralized consensus mechanism. Tackling this challenge in a low-latency， high-throughput system is particularly challenging and a key focus of Unit-e’s design.

　　低延迟。延迟对于面向消费者的支付系统至关重要，特别是对于销售时点情报系统（POS）。因此，Unit-e旨在实现链上交易的15秒，及脱链交易的2-4秒的确认延迟。尽管一些加密货币今天实现了类似速度的延迟，但它们是以牺牲去中心性为代价的。我们发现这种权衡并不是根本的；使用去中心化算法实现快速确认是可能的。我们彻底地重新设计共识机制的以及支付渠道网络的新路由算法，避开了这一挑战。

　　Low Latency.Latencyis critical in a consumer-facing payment system， particularly for point-of-sale transactions. Unit-e therefore aims to achieve confirmation latencies on theorder of 15 seconds for on-chain transactions， and 2-4 seconds for off-chain transactions. Although some cryptocurrencies achieve comparable latencies today， they do so at the expense of decentralization. We find that this tradeoff is not fundamental； it is possible to achieve fast confirmation using decentralized algorithms. We sidestep this challenge through a ground-up redesign of consensus mechanisms， coupled with new routing algorithms for payment channel networks.

　　在共识方面，我们开发了强大理论性能保证的新算法，这些算法能够接近区块链中物理可能的极限。我们设计此类算法的过程，包括仔细和系统地将区块链解构为其核心功能，并从头开始重建。关键的直观是，为了获得良好的吞吐量和延迟，提议和确认区块的过程应该在算法上分开。我们在第3章中讨论了安全有效的共识算法的设计。

　　On the consensus front， we develop new algorithms with strong theoretical performance guarantees， which are able to approach the limits of what is physically possible in a blockchain. Our process for designing such algorithms involve scarefully and systematically deconstructing blockchains into their core functionalities and rebuilding them from scratch. A key intuition is that in order to get good throughput and latency， the processes of proposing and confirming blocks should be algorithmically separated. Our design of secure and efficient consensus algorithms is discussed in Chapter 3.

　　与快速的链上权益证明（PoS）共识并行，我们实施了针对大量低价交易（例如POS）的支付渠道网络。支付渠道网络是覆盖网络，使用链上共识来在用户对之间建立托管账户（或通道）。利用聪明的加密结构，用户可以这些通道的路径发送交易，即使这两个端点本身不共享通道。重要的好处在于用户可以在不等待区块链确认的情况下即时验证交易。与链上交易相比，这显著减少了确认延迟；主要的延迟将源于将交易传递给接收者，对于直接通道可能只需要一秒钟，这是一种快速的点对点操作。

　　In parallel with fast on-chain proof-of-stake （PoS） consensus， we implement a payment channel network targeted towards high-volume， low-value transactions （e.g.point-of-sale transactions）. Payment channel networks are overlay networks that use on-chain consensus to set up escrow accounts （or channels） between pairs of users. By exploiting clever cryptographic constructs， users can route transactions over a path of these channels， even if the two endpoints do not share a channel themselves. The key benefit is that users can verify transactions instantaneously without waiting for confirmation from the blockchain. This significantly reduces confirmation latency compared to on-chain transactions； the main delay stems from passing the transaction to the recipient， which is a fast， point-to-point operation that can take as little as a second for direct channels.

　　吞吐量。与延迟密切相关的概念是吞吐量，即每秒处理的交易数。我们的目标是每秒5,000-10,000笔交易。做个比较，请注意，Visa的网络平均每秒处理近1,700笔交易，在峰值时处理的数量级更大。同样为了比较，比特币当前的平均吞吐量估计在每秒3.3-7笔交易，而以太坊每秒10-30笔。弥合这一巨大差距在技术上是非凡的，需要大量创新。另外，我们注意到目标吞吐量指标已达到现代代表性点对点（P2P）网络的物理极限；20Mbps的网络在不做出严重妥协的情况下（通常在安全性方面），物理上物理上无法无法每秒处理更多交易。

　　Throughput. A closely-related concept to latency is throughput—the number of transactions processed per second. We are targeting throughputs of 5,000-10,000 transactions per second. For comparison， note that Visa’s networks process almost 1,700 transactions per second on average， and an order of magnitude more at its peak. Also for comparison， Bitcoin’s current average throughput is estimated between 3.3 and 7 transactions per second， and Ethereum reaches between 10-30 transactions per second. Bridging this large gap is technically nontrivial andrequires significant innovation. As an aside， we note that the target throughput metrics are already at the physical limits of a typical modern P2P network； a 20 Mbps network physically cannot handle substantially more transactions per second without making severe compromises （typically insecurity）.

　　为了实现我们的目标吞吐数，我们依赖于新共识机制、全新分片方式和支付渠道网络的组合。我们设计了新的共识算法（第3章）和新的P2P网络算法，后者与共识算法的信息要求同步（第4章）。这使我们在链上交易时，能仅受限于网络层，实现最佳吞吐量（并同时实现最佳延迟和安全性）。我们提出了基于对区块链数据进行编码的新分片算法，这些算法在安全性，存储和计算方面实现了最佳权衡；这是第5章的工作。最后，我们在第6章中设计了新的支付渠道网络，我们提出了一种新的路由算法，与最先进的路由提议相比，可以实现多出50％的交易吞吐量，而不在延迟方面有所牺牲。

　　To achieveour target throughput figures， we rely on a combination of novel consensus mechanisms， entirely new ways of sharding， and payment channel networks. We design novel consensus algorithms （Chapter 3） and new peer-to-peer （P2P）networking algorithms that breathe in sync with the informational imperatives of the consensus algorithms （Chapter 4）. This allows us to achieve optimal throughput （and latency and security， simultaneously）， constrained only by the physical limits of the networking layer， for on-chain transactions. We propose new sharding algorithms based on coding the blockchain data which achieve optimal tradeoffs in security， storage， and computation； this is conducted in Chapter 5. Finally， we design new payment channel networks in Chapter 6, wherewe propose a new routing algorithm that achieves up to 50％ higher transactionthroughput than state-of-the-art routing proposals， without sacrificing latency.

　　可用性。可用性对于支付系统至关重要，因为顾客必须能随时支付。原则上，Unit-e节点的分布式P2P网络提供了一些防止随机网络波动的保护。实践中，比特币近十年来一直非常可靠；2011年以来，测量结果显示网络可靠性超过99.9999％。第7章中讨论的Unit-e激励机制旨在激励和奖励参与节点以达到相似的可用性水平。

　　Usability. Usability is critical to a payment system， since customers must be able to make payments whenever they wish. In principle， the distributed， peer-to-peer network of Unit-e nodes gives some protection against random network fluctuations. In practice， Bitcoin has been extraordinarily reliable over its existence for nearly a decade； measurements have shown anetwork reliability exceeding 99.9999％ since 2011. The incentive mechanisms of Unit-e —discussed in Chapter 7— are designed to motivate and reward participant nodes towards similar levels of availability.

　　隐私。隐私是将加密货币用作真实支付系统的主要挑战之一。按照设计，加密货币被设计为透明的；区块链本质上是一个公开的，可核实的交易记录。但是，如果人们要将加密货币用于日常交易，这可能会导致严重的隐私侵犯。为了实现这种权衡，Unit-e将整合隐私保护以防止区块链攻击（例如，Zcash，一种保护隐私的加密货币，提供了我们设计的起点）。Unit-e还通过提出一种名为蒲公英（Dandelion）的新型隐私解决方案来防范网络级攻击，该解决方案可以防止网络攻击者将用户的交易与其IP地址联系起来（即使数据被加密）。

　　Privacy. Privacy is one of the primary challenges associated with using a cryptocurrency as a real payment system. By design， cryptocurrencies aredesigned to be transparent； the blockchain is inherently a public， verifiablerecord of transactions. However， this can lead to significant privacy violations if people are to use the cryptocurrency for everyday transactions. To navigate this tradeoff， Unit-e will incorporate privacy protections against blockchain-level attacks （e.g.， Zcash， a privacy-preserving cryptocurrency， provides a starting point of our design）. Unit-e also protects against network-level attacks by proposing a novel privacy solution called Dandelion that protects against network adversaries linking users’ transactions to their IP addresses （even with the data being encrypted）.

　　1.2概要 Outline

　　我们的目标是采用第一性原则，全栈方法来设计Unit-e的支付系统。在这份宣言中，我们为正式提出的关键要求提供了一套全面的解决方案，并为区块链的持续研究奠定了基础。鉴于主题广泛，我们自然地将解决方案划分为几个章节的篇幅。表1.1总结了性能指数与不同章节之间的联系。这些章节涵盖了在信息理论、编码理论、通信理论、分布式算法、经济学和数据网等多学科所获得的集体经验和知识下进行的区块链的最新研究。接下来提供所有章节的主要发现的摘要。

　　Our goal is to take a first-principles， full-stack approach to designing Unit-e’s payment system. In this manifesto， we provide a comprehensive set of solutions to the key requirements set up formally earlier， as well as setting the stage for continued research on blockchains. Given the vast subject matter we have found it natural to divide the solution space into several chapters. Table 1.1 summarizes the connection between the figures of merit and the different chapters. The chapters cover state of the art researchin blockchains conducted under the collective experience and knowledge gained by the diverse disciplines of information theory， coding theory， communication theory， distributed algorithms， economics， and data networking. A summary of the main findings of all the chapters is provided next.

　　第2章 Unit-e：设计概要 Unit-e: Summary of Design

　　本节的目的是总结Unit-e的设计和研究贡献。虽然前一节概述了我们的技术要求，但我们以加密货币架构组件的抽象化开始本节。本宣言的每一章都将以第一性原理、基础设计来解决其的一个部分。

　　The goal of this section is to summarize the design and research contributions of Unit-e. Whereas the previous section outlined our technical requirements， we begin this section with an abstraction of the architectural components of a cryptocurrency. Each chapter of this manifesto will address one piece of this abstraction through a first-principles， ground-up design.

　　2.1 加密货币的架构TheArchitecture of Cryptocurrencies

　　区块链通常由分层模型在概念上表示，非常类似于OSI网络模型。在设计Unit-e时，我们使用这个基于图层的模型作为起点，并将其扩展，以包含与加密货币尤其相关的组件。如图2.1所示，第1层技术指的是核心区块链；这包括从共识机制到数据结构到网络堆栈的所有内容。传统上，区块链开发和研究的大部分已经解决了第1层。而第2层描述了使用底层区块链来构建应用程序的技术。虽然没有第1层的情况下，第2层技术也不能存在，但某些第2层技术可能对区块链的长期可行性和扩展性至关重要。支付渠道网络，如比特币的Lightning网络、以太坊的Raiden网络，是著名的范例技术。

　　Blockchains are often represented conceptually by a layered model， much like the OSI networking model. In designing Unit-e， we use this layer-based model as a starting point， and expand it to include components that are specifically relevant to cryptocurrencies. As shown in Figure 2.1, Layer 1 technologies refer to the core blockchain； this includes everything from consensus mechanisms to datastructures to the networking stack. Traditionally， the bulk of blockchain development and research has addressed layer 1. Layer 2 instead describes technologies that use an underlying blockchain to build applications. Although layer 2 technologies cannot exist without layer 1, certain layer 2 technologies could prove essential for the long-term viability and scalability of blockchains. Payment channel networks—such as Bitcoin’s Lightningnetwork and Ethereum’s Raiden network—are prominent exemplar technologies.

　　层的抽象化对于思考区块链的开发和结构很有用。但是，加密货币的某些方面跨越多层且不易分类。隐私、经济学和治理就是其中的三个。隐私是指，用户在不向其他用户透露有关此交易的信息的情况下进行交易的能力。特别地，金融系统具有严格的隐私要求，最初的区块链设计不一定能满足这些要求。此外，在第1层和第2层都存在隐私要求，因此，层模型并不自然地包含隐私技术。因此，我们在图2.1中将隐私视为包括两层。经济学是指，激励用户提供日常运营所需的存储、计算和带宽的机制。它还指，加密货币和法币间不稳固的关系，以及如何相互计价。由于经济学严格依赖于关于第1层、第2层及隐私层的算法决策，我们认为经济学包括它们所有。最后，治理是指，制定网络决策的过程和规则，从技术建议到灾难恢复。治理可以影响区块链的所有方面，因此包含所有先前的层。虽然治理对任何项目的成功至关重要，但它不是传统意义上的研究课题，因此我们不会在本研究宣言中进一步讨论。

　　The layer abstraction is useful for reasoning about the development and structure of blockchains. However， some aspects of cryptocurrencies span multiple layers and are not easily categorized. Three such aspects are privacy， economics， and governance. Privacy refers to the ability of users to make transactions without revealing information about this transaction to other users. Financial systems in particular have stringent privacy requirements， which are not necessarily satisfied by naive blockchain designs. Moreover， privacy requirements exist atboth layers 1 and 2, so privacy technologies are not naturally captured by the layer model. Hence we view privacy as encompassing both layers in Figure 2.1.Economics refers to the mechanisms that incentivize users to offer storage， computation， and bandwidth needed for daily operations. It also refers to the tenuous relation between cryptocurrencies and fiat currencies， and how to priceone in terms of the other. Since economics depend closely on algorithmic decisions regarding layers 1 and 2, as well as the privacy layer， we think ofeconomics as encompassing all of them. Finally， governance refers to the processes and rules for making decisions about the network， ranging from technical recommendations to disaster recovery. Governance can affect all aspects of the blockchain， and therefore encompasses all of the previouscomponents. Although governance is critical to the success of any project， itis not a research topic in the traditional sense， and hence we do not discussit further in this research manifesto.

　　2.2 Unit-e的设计 Unit-e’s design

　　有了这种抽象化，我们就可系统性地处理架构的每个组成部分。宣言从核心第1层技术开始：共识、存储和计算。接下来，我们将介绍第2层的支付渠道网络，其次是经济学和隐私。对于这些技术中的每一种，我们首先考虑根据物理定律下可能的性能水平。 接下来，我们提出量身定制的算法以达到这些物理限制。在下文中，我们总结了每章的内容和研究贡献。

　　Armed with this abstraction， we systematically tackle each component of the architecture. Our manifesto begins with the core layer 1 technologies: consensus， storage， and computation. Next， we cover payment channel networks at layer 2, followed by economics and privacy. For each of these technologies， we first consider what performance levels are possible according to the laws of physics. Next， we propose algorithms that are tailored to meet those physical limits. In the following， we summarize the contents and research contributions of each ofthese chapters.

　　第3章：共识 Consensus

　　对于给定的安全级别，两个关键性能指标代表了区块链共识协议：吞吐量和延迟。在去中心设定下，这些措施受到底层物理网络属性的限制，即通信容量和光速传播延迟。我们阐述了现有加密货币远未达到这些物理限制，开始本章。自然会有一个疑问，是否存在一共识算法能达到接近物理极限的性能指标。我们提出了Prism，给出了肯定的答案，棱镜（Prism）是一种新的区块链协议，可以实现：1）在多达50％的对抗性节点下的安全性；2）高达网络容量C的最大吞吐量；3）与传播延迟D成比例的诚实交易的最优序确认延迟，伴随带宽延迟乘积CD中指数性小的确认错误概率；4）所有交易的最终总排序。我们设计此协议的方法是基于将区块链解构为其基本功能，并系统地扩展这些功能以接近其物理极限。我们首先在工作量证明设置中提出Prism，然后讨论如何将其底层的直观扩展到权益证明共识算法。

　　For a given security level， two key performance metrics characterize a blockchain consensusprotocol: throughput and latency. In a decentralized setting， these measuresare limited by underlying physical network attributes—namely， communication capacity and speed-of-light propagation delay. We begin this chapter by showing that existing cryptocurrencies operate far from these physical limits. A natural question is whether any consensus algorithm can achieve performance metrics close to the physical limits. We answer this question in the affirmative by presenting Prism， a new blockchain protocol that achieves 1） security against up to 50％ adversarial nodes； 2） optimal throughputup to the capacity C of the network； 3） order-optimal confirmation latency for honest transactions proportional to the propagation delay D， with confirmation error probability exponentially small in the bandwidth-delay product CD； 4）eventual total ordering of all transactions. Our approach to the design of this protocol is based on deconstructing the blockchain into its basic functionalities and systematically scaling up these functionalities to approachtheir physical limits. We begin by presenting Prism in the proof-of-work setting， then discuss how to extend the intuitions underlying it to a proof-of-stake consensus algorithm.

　　Prism建立在以下直观上：为了实现高吞吐量，系统应经常产生区块。然而，天真地增加区块产量将导致分叉，这会阻碍交易确认并降低协议的安全性。为了应对这种张力，我们在架构上将产生新区块的行为与确认它们的行为分开。其他系统已经提出解耦共识的各个方面。然而，Prism特殊的解耦能带来重度结构化的有向无环图（DAG），它能同时促进分析，同时提供最佳延迟和吞吐量保证。该DAG如图2.2所示。

　　Prism builds on the following intuition: to achieve high throughput， a system should produce blocks frequently. However， naively increasing the block production rate leads to forking， which hinders transaction confirmation and reduces the security ofthe protocol. To resolve this tension， we architecturally separate the act of producing new blocks from the act of confirming them. Other systems have proposed decoupling various aspects of consensus. However， Prism’s decoupling in particular leads to aheavily-structured directed acyclic block-graph （DAG） that simultaneously facilitates analysis， while giving optimal latency and throughput guarantees.This DAG is illustrated in Figure 2.2.

　　第4章：网络意识共识 Network-Aware Consensus

　　在第3章中，我们提出了一种新的共识协议，在给定网络中达到物理限制。这里，“网络”是指整个网络堆栈，从硬件到拓扑到中继协议。在第3章中，整个堆栈用最坏情况区块延迟的简单模型表示，如3.3节所述。该模型在共识的文献中是相当标准的，但它抽象去了现实网络中出现延迟的大部分细微差别。在第4章中，我们思考了一个更详细、随机、以真实加密货币的测量得出的网络模型。借助这种更复杂的网络模型，我们提出了一种新的网络协议，可以从一大类协议中提高共识算法的吞吐量，包括中本共识和Prism。该协议称为梭鱼（Barracuda），要求每个提议者节点在提议前，轮询其同侪以获取新区块，并更新其区块树的本地视图。这个看似简单的轮询过程为提议者提供了有关全局区块树的更多信息，从而降低了提议者的新区块导致分叉的可能性。事实上，我们证明了，如果每个提议者在提议前轮询e节点，对于小值e，最终效果就好像整个网络的平均区块延迟提速了e因子。值得注意的是，此优势没有任何硬件升级或网络的其他重大变化。因此，Barracuda是一个有用的开端，用于改善区块链吞吐量和延迟，可与第3章中的共识算法改进并行。

　　In Chapter 3, we proposed a new consensus protocol to meet the physical limits of a fixed network. Here， by ‘network’， we mean the entire networking stack， ranging from the hardware to the topology to the relaying protocols. In Chapter 3, this whole stack is represented with a simple model for worst-case block delay， described in Section 3.3. This model is fairly standard in the consensus literature， but it abstracts away most of the nuances that characterize delays in real networks. In Chapter 4, we consider a more detailed， random network model informed by measurements of real cryptocurrencies. Armed with this more sophisticated network model， we propose a new networking protocol that improves throughput for any consensus algorithm from a broad class of protocols， including Nakamoto consensus and Prism. This protocol， called Barracuda， requires every proposer node to poll its peers for new blocksand update its local view of the blocktree before proposing a new block. This deceptively simple polling procedure gives the proposer more information about the global blocktree， thereby reducing the probability of the proposer’s new block causing a fork. In fact， we show that if each proposer polls e nodes before proposing， for small values of e， theend effect is the same as if the entire network had been a factor of e faster in mean block delay. Notably， this benefit comes without any hardware upgrades or other substantial changes to the network. Barracuda is therefore a useful primitive for improving blockchain throughput and latency that operates in parallel to the consensus algorithm improvements from Chapter 3.

　　第5章：可扩展存储和计算 Scalable Storage and Computation

　　第3章介绍了如何处理延迟和吞吐量的物理限制。在第5章中，我们将重点放在接近存储和计算效率的物理限制上。存储和计算容量一个微小的上限是节点数量线性扩展的上限。更简单地说，节点数量加倍也应该使系统可以存储的数据量和可以执行的计算量加倍。目前的区块链设计远未达到这一基本限制；实际上，在现有系统中，更多用户参与将导致较低的存储和计算效率。这种反向扩展性，部分是因为加密货币的全球要求：所有（或许多）用户都希望存储和处理区块链，因此产生的存储和计算成本随着时间和用户参与而增加。目前提高扩展性的建议（称为分片）通过提高存储和计算效率来解决这一问题；然而，它们的代价通常是安全。系统中分片越多，攻击者就越容易超越单个分片。和以前一样，自然会有疑问，是否可以同时实现最佳存储效率，计算效率和安全性。

　　Chapter 3 showed how to approach the physical limits on latency and throughput. In Chapter 5, we focus on approaching physical limits on storage and computation efficiency. A trivial upper bound is one where storage and computational capacity scale linearly in the number of nodes. Put more simply， doubling the number of nodes should also double the amount of data the system can store and computation it can execute. Current blockchain designs operate far from this fundamental limit； indeed， in existing systems， more user participation actually leads to lower storage and computational efficiency. This reverse scalability arises in part because of the global requirements of cryptocurrencies: all （or many） users are expected to store and process the blockchain， so the resulting storage and computational costs increase with time and userparticipation. Current proposals for improving scalability （called sharding） address this tradeoff by improving storage and computational efficiency； however， they typically do so at the cost of security. The more shards exist ina system， the easier it becomes for an adversary to overtake any single shard. As before， a natural question is whether one can simultaneously achieve optimal storage efficiency， computational efficiency， and security.

　　在本章中，我们提出PolyShard（多片），来给出肯定答案，PolyShard是一种存储和计算解决方案，在不牺牲安全性的情况下，更多用户能达到更高效率。PolyShard将来自编码存储和编码计算的经典思想与拉格朗日编码计算的最新突破性成果相结合，以规避大多数分片系统表现出的权衡。关键的直观是，它以一种允许完美恢复的方式混合来自不同用户和交易的数据，如图2.3所示。值得注意的是，PolyShard同时实现了安全性，存储和计算成本之间的最佳权衡，在物理限制的一个小常数因子下运行。

　　In this chapter， we answer this question affirmatively by presenting PolyShard， a storage and computation solution that grows more efficient with more users without sacrificing security. PolyShard combines classical ideas from coded storage and coded computation with recent breakthrough results on Lagrange-coded computing to circumvent the tradeoffs that characterize most sharding systems. The key intuition is that it mixes up data from different users and transactions in a way that allows perfect recovery， illustrated intuitively in Figure 2.3. Notably， PolyShard simultaneously achieves an optimal tradeoff between security， storage， and computational costs， performing within a small constant factor of the limits imposed by physics.

　　第6章：支付渠道网络 Payment Channel Networks

　　在第3章和第5章中，我们提出了第1层算法，它们达致吞吐量、延迟、存储、计算和安全性的物理限制。这些物理限制特别适用于分布式共识系统（即第1层）。在某些情况下，第2层的扩展性解决方案受到的物理限制更宽松；如果正确使用，这可以进一步提高效率。

　　In Chapters 3and 5 we propose layer 1 algorithms that meet physical limits on throughput， latency， storage， computation， and security. These physical limits apply specifically to distributed consensus systems （i.e.， layer 1）. In some cases， layer 2 scalability solutions are subject to more lenient physical constraints； this can lead to further efficiency gains if harnessed properly.

　　一个值得注意的例子是支付渠道网络（PCN）。PCN消除每个交易中的区块链涉及需要，提高了加密货币的扩展性；为实现这一目标，用户提前在区块链上托管资金，并使用加密数据结构以便在以后快速有效地从这些托管中提取钱。PCN的示例包括比特币中的Lightning网络、以太坊中的Raiden网络。尽管有这些承诺，支付网络仍然存在许多技术挑战，这些挑战尚未得到研究界的广泛关注。特别是，与更成熟的网络（如互联网或数据中心）中的路由算法相比，现在PCN中使用的路由算法是不堪一击的。

　　A notable example is payment channel networks （PCNs）. PCNs improve the scalability of cryptocurrencies by removing the need to involve the blockchain with every transaction； to achieve this， users escrow money on the blockchain ahead of time， and use cryptographic data structures to quickly and efficiently draw from these escrows later. Examples of PCNs include the Lightning network in Bitcoin and the Raiden network in Ethereum. Despite their promise， payment networks presents numerous technical challenges that have yet to receive much attention from the research community. In particular， the routing algorithms used in today’s PCNs are naive compared to routing inmore mature networks， like the Internet or data centers.

　　在第6章中，我们提出Spider（蜘蛛），这是一种针对PCN的网络解决方案，它使用了来自网络文献中经验证的思想。Spider使用分组交换、不平衡感知的路由算法来降低PCN内的成本和重平衡频率。换句话说，它减少了必须托管的金额，这些钱是用来支持给定交易负载的。我们发现，对于稳定的交易工作流程，一个小因子下，Spider可以实现最佳吞吐量。模拟时，与目前最先进的算法比，Spider的交易吞吐量提高了50％，并且不会牺牲交易处理延迟。

　　In Chapter 6 we present Spider， a networking solution for PCNs that exploits tried-and-tested ideas from the networking literature. Spider uses packet-switched， imbalance-aware routing to reduce costs and rebalancing frequency within a PCN. In other words， it reduces the amount of money that must be escrowed to support a given transaction load. We find that for stable transaction workflows Spider achieves within a small factor of optimal throughput. In simulation， Spider achieves up to 50％ higher transaction throughput compared to state-of-the-art algorithms， without sacrificing transaction processing delay.

　　第7章：经济学 Economics

　　经济学是加密货币设计中一个核心的开放性问题：加密货币的价值是什么？我们应该如何设计奖励机制来激励理性行动者参与区块链系统？这些问题通常以特例假设（ad-hoc）的方式回答，主要是由哲学或质化论证所驱动。第7章的目标是提供一个量化回答这些问题的框架。

　　Economics are a central， open question in the design of cryptocurrencies: what is the value of a cryptocurrency？ How should we design reward mechanisms to incentivize rational players to participate in blockchain systems？ These questions are typically answered in an ad-hoc manner， motivated primarily by philosophical or qualitative arguments. The goal of Chapter 7 is to provide a framework for answering these questions quantitatively.

　　一种常见的观点是，加密货币应根据其功能定价。因此，像Unit-e这样的支付通证应根据它处理的交易的法币价值定价。然而，将这种高层次的思想转化为精确的估值是一项挑战；尤其是，通证价值与难以确定的通证速度相关联。此外，现有模型主要涵盖工作量证明加密货币，其高硬件和电力成本实质上影响网络的价值。这些模型无法延伸到物理成本小得多的权益证明通证。

　　A common view is that cryptocurrencies should be valued according to their functionality. So a payment token like Unit-e should be valued according to the fiat value of transactions it processes. However， translating this high-level thought into aprecise valuation is challenging； typically， token values are tied to token velocity， which is difficult to pin down. Moreover， existing models have primarily covered proof-of-work cryptocurrencies， where the high hardware and electricity costs materially affect the value of the network. These models do not extend to proof-of-stake tokens， where physical costs are much smaller.

　　在本章中，我们设计了为权益证明支付通证定价的量化模型，基于传统资产评估技术。这些模型依赖于一关键观察：在权益证明加密货币中，用户存储通证以参与保护网络的共识协议。作为参与这些协议的补偿，他们以交易费用的形式获得奖励。因此，单个存储的通证价值是由未来奖励的累积确定的，这是所有者从共识机制接收的。假设交易费用是交易价值的固定百分比，我们就可以将整体通证价值与全网交易的法币值联系起来。

　　In this chapter， we design quantitative models for valuing proof-of-stake payment tokens based on conventional asset valuation techniques. These models rely on one key observation: in proof-of-stake cryptocurrencies， users deposit tokens to participate in consensus protocols that secure the network. As compensation for participating in these protocols， they receive rewards in the form of transaction fees. Thus， the value of a single deposited token is determined by the cumulative future rewards the owner will receive from the consensus mechanism. By assuming transaction fees are a fixed percentage of transaction value， we can tie the overall token value to the fiat value of transactions executed on the network.

　　第8章：隐私和身份管理 Privacy and Identity Management

　　身份概念对金融交易至关重要，不论是顾客还是商家。对加密货币的一个重要观察就是，支付系统不需要知道参与者的身份；这类似于现金和物物交易系统的经济。在本章中，我们将探讨在相关双方（即商家和客户）间维持强有力的身份管理所需的基本挑战，同时减少向中间人泄漏不必要的信息。我们将分两部分研究这些挑战：第一部分涉及网络参与者（如商家和客户）间的身份管理。第二部分涉及技术上更加困难的区块链中的身份管理问题。后一方面是第8章的重点；我们在区块链级别（即防止交易泄漏）上研究隐私，也在网络级别（防止交易和IP地址间的联系）研究隐私。

　　The notion of identity is critical to financial transactions， both to customers and merchants. A key observation of cryptocurrencies is that payment systems do not need to know the identities of their participants； this is similar to cash and barter system economies. In this chapter we explore the fundamental challenges arising from the need to maintain strong identity management between relevant parties （i.e.， merchants and customers）， while cutting out unnecessary information leakage to the middleman. We study these challenges in two parts: the first part involves identity management among network participants， such asmerchants and clients. The second part deals with the technically much harder problem of identity management in the blockchain. This latter aspect is the specificfocus of Chapter 8； we study privacy both at the blockchain level （i.e.preventing transaction leakage） and at the network level （preventing linkagebetween transactions and IP addresses）.

　　我们描述了使用零知识技术提供区块链数据隐私的现有技术，开始本章。我们总结了主要的扩展性挑战，并概述了不同方法之间的权衡，包括zk-SNARKS和zk-STARKS。接下来，我们将讨论网络级攻击的情况，包括现有攻击和分析的摘要。最后，我们提出了一个名为Dandelion（蒲公英）的网络级隐私解决方案。Dandelion阻止网络攻击者将交易链接到IP地址；直观地说，为实现这一点，它将交易在P2P网络中继改变为非对称传播模式的方式，如图2.4所示。我们展示了Dandelion以低延迟成本实现了最佳隐私保证，并讨论了在低延迟加密货币（如Unit-e）中实施Dandelion的实际考虑因素。

　　We start the chapter by describing existing techniques for providing blockchain data privacyusing zero-knowledge techniques. We summarize primary scalability challenges， and outline the tradeoffs between different approaches， including zk-SNARKS and zk-STARKS. Next， we discuss the landscape of network-level attacks， including a summary of existing attacks and analysis. Finally， we present a proposed solution for network-level privacy called Dandelion. Dandelion prevents network adversaries from linking transactions to IP addresses； intuitively， it achievesthis by changing the way that transactions are relayed over the P2P network to an asymmetric spreading pattern， shown in Figure 2.4. We show that Dandelion achieves optimal privacy guarantees at low latency cost， and discuss practical considerations for implementing Dandelion in a low-latency cryptocurrency like Unit-e.