bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 8쪽 중 1쪽
Celer Network: Bring Internet Scale to Every Blockchain
ScaleSphere Foundation Ltd. (“Foundation”)
June 15, 2018
Draft, subject to change.
Abstract. Just like how the 56Kbps dialup Internet in the 90s cannot possibly support
4K video streaming, the insucient scalability of today’s blockchain is the key factor
limiting its use cases. Current blockchains have low throughput because each operation
needs to be processed by the vast majority of nodes to reach on-chain consensus, which
is exactly ”how to build a super slow distribution system”. Ironically, the on-chain
consensus scheme also leads to poor privacy as any node can see the full transaction
history of one another. While new consensus algorithms keep getting proposed and
developed, it is hard to free on-chain consensus from its fundamental limitations.
O↵-chain scaling techniques allow mutually distrustful parties to execute a contract
locally among themselves instead of on the global blockchain. Parties involved in the
transaction maintain a multi-signature fraud-proof o↵-chain replicated state machine,
and only resort to on-chain consensus when absolutely necessary (e.g., when two par-
ties disagree on a state). O↵-chain scaling is the only way to support fully scale-out
decentralized applications (”dApps”) with better privacy and no compromise on the
trust and decentralization guarantees. It is the inflection point for blockchain mass
adoption, and will be the engine behind all scalable dApps.
Celer Network is an Internet-scale, trust-free, and privacy-preserving platform where
everyone can quickly build, operate, and use highly scalable dApps. It is not a stan-
dalone blockchain but a networked system running on top of existing and future
blockchains. It provides unprecedented performance and flexibility through innova-
tion in o↵-chain scaling techniques and incentive-aligned cryptoeconomics.
Celer Network embraces a layered architecture with clean abstractions that enable
rapid evolution of each individual component, including a generalized state channel
and sidechain suite that supports fast and generic o↵-chain state transitions; a provably
optimal value transfer routing mechanism that achieves an order of magnitude higher
throughput compared to state-of-the-art solutions; a powerful development framework
and runtime for o↵-chain applications; and a new cryptoeconomic model that provides
network e↵ect, stable liquidity, and high availability for the o↵-chain ecosystem.
IMPORTANT: YOU MUST READ THE FOLLOWING DISCLAIMER IN FULL BEFORE
CONTINUING
The sale (“Token Sale”) of CELR (“Tokens”), the cryotographic token associated with the Celer
Network as detailed in this whitepaper (“Whitepaper”), is only intended for, made to or directed at,
only certain persons. This Whitepaper is not a prospectus or offer document of any sort and is not
intended to constitute an offer of securities of any form, units in a business trust, units in a collective
investment scheme or any other form of investment, or a solicitation for any form of investment in
any jurisdiction. No regulatory authority has examined or approved of any of the information set out
in this Whitepaper. This Whitepaper has not been registered with any regulatory authority in any
jurisdiction.
By accessing and/or accepting possession of any information in this Whitepaper or such part thereof
(as the case may be), you represent and warrant to cSpeed Ltd (BVI Business Company Number
1976167) (the “Token Vendor”) that:
(a) you are not an Excluded Person (as defined below);
(b) you are not located in a jurisdiction where the Token Sale is prohibited, restrict ed or
unauthorised in any form or manner whether in full or in part under its laws, regulatory
requirements or rules;
(c) you have read the entirety of this Whitepaper and understand the risks entailed in your
purchase of the Tokens;
(d) you agree to be bound by the limitations and restrictions described herein; and
(e) you acknowledge that this Whitepaper has been prepared for delivery to you so as to assist
you in making a decision as to whether to purchase the Tokens, and is purely for reference
without any intention to create legal relations between you and the Token Vendor, or to be
legally binding or enforceable by you against the Token Vendor.
Please do not proceed further into the contents of this Whitepaper if you do not agree with any of the
above. Please refer to the section entitled “Important Notice” and the sections Entitled “Disclaimer
Of Liability”, “No Representations And Warranties”, “Representations And Warranties By You”,
“Cautionary Note On Forward-Looking Statements”, “Third Party Information And No Consent Of
Other Persons”, “Terms Used”, “No Advice”, “No Further Information Or Update”, “Restrictions On
Distribution And Dissemination”, “No Offer Of Investment Or Registration” and “Risks And
Uncertainties” carefully before proceeding further with this Whitepaper.
Contents
1 Introduction 4
1.1 Celer Technology Stack . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Celer Cryptoeconomics . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2 cChannel: the F oundation of O↵-Chain Scaling 10
2.1 Generalized State Channel . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.1 Key Idea and a Simple Example . . . . . . . . . . . . . . . . . 10
2.1.2 Design Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.3 General Specification . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.4 Common Utilities . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.1.5 Out-of-box Features . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 Alternative Channel Model with Sidechains . . . . . . . . . . . . . . . 19
3 cRoute: Provably-Optimal V alue T ransfer Routing 21
3.1 Challenges in State Channel Network Routing . . . . . . . . . . . . . . 21
3.2 Distributed Balanced Routing (DBR) . . . . . . . . . . . . . . . . . . 23
3.2.1 System Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.2 Protocol Description . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.3 Throughput Performance of DBR . . . . . . . . . . . . . . . . . 27
3.3 Discussions of DBR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3.1 Failure Resilience . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3.2 Privacy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.4 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4 cOS: O↵-chain Decentralized Application Operating System 35
4.1 Directed Acyclic Graph of Conditionally Dependent States . . . . . . 35
4.2 O↵-chain Application Development Framework . . . . . . . . . . . . . 36
4.3 O↵-chain Application Runtime . . . . . . . . . . . . . . . . . . . . . . 38
5 cEconomy: O↵-chain Cryptoeconomics Mechanism Design 40
5.1 Tradeo↵s in O↵-chain Ecosystems . . . . . . . . . . . . . . . . . . . . . 41
5.1.1 O↵-Chain Scalability vs. Liquidity . . . . . . . . . . . . . . . . 41
5.1.2 O↵-Chain Scalability vs. Availability . . . . . . . . . . . . . . . 42
2
5.2 cEconomy Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.2.1 Proof of Liquidity Commitment (PoLC) Mining . . . . . . . . . 44
5.2.2 Liquidity Backing Auction (LiBA) . . . . . . . . . . . . . . . . 45
5.2.3 State Guardian Network . . . . . . . . . . . . . . . . . . . . . . 49
5.2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
6 Conclusion 52
7 F ounding T eam 53
3
1. Introduction
Many modern economic activities are essentially the flow and exchange of information
and value. Over the past two centuries, the transfer of information has evolved from
discrete events through pigeon networks to continuous flows through the speed-of-light
Internet. However, the value transfer portion is far from light speed and is still very
much discrete events controlled by segregated financial silos. This mismatch creates a
devastating bottleneck in economic evolution: no matter how fast information flows,
the expensive and slow value transaction is limiting the productive exchange of the
two.
Essentially a revolutionary abstraction of trust among distrustful parties that re-
sults in an incentive-aligned distributed consensus, blockchain technology o↵ers the
foundation to dismantle segregated financial silos and dramatically expand the scope
and freedom of global value flows. In practice, however, blockchain is deviating further
away from the speed-of-light vision due to its low processing power compared to tra-
ditional value transfer tools. Scalability is a fundamental challenge that is hindering
mass adoption of blockchain technology.
We envision a future with decentralized ecosystems where people, computers, mo-
bile and Internet-of-Things (”IOT”) devices can perform secure, private, and trust-free
information-value exchange on a massive scale. To achieve this, blockchains should
match the scale of the Internet and support hundreds of millions or billions of trans-
actions per second. However, given the processing speed of existing blockchains (i.e.,
a few or tens of transactions per second), is it really possible to bring the scale of the
Internet to blockchains? The answer is yes but only with o↵-chain scaling.
While on-chain consensus is the foundation of blockchain technology, its limitations
are also obvious. In a sense, consensus is the opposite of scalability. For any distributed
system, if all nodes need to reach consensus on every single transaction, its perfor-
mance will be no better (in fact, much worse due to communication overhead) than a
centralized system with a single node that processes every transaction, which means
the system is eventually bottlenecked by the processing power of the slowest node.
On-chain consensus also has severe implications on privacy, because all transactions
are permanently public. A few on-chain consensus improvements have been proposed
including sharding and various Proof-of-X mechanisms. They make the blockchain rel-
atively faster with di↵erent tradeo↵s in performance, decentralization, security, and
4
finality, but cannot change the fundamental limitations of on-chain consensus.
To enable Internet-scale blockchain systems with better privacy and no compromise
on trust or decentralization, we have to look beyond on-chain consensus improvements.
The core principle to design a scalable distributed system is to make operations on
di↵erent nodes mostly independent. This simple insight shows that the only way to
fully scale out decentralized applications is to bring most of the transactions o↵-chain,
avoid on-chain consensus as much as possible and use as a last resort. Related tech-
niques include state channel, sidechain, and o↵-chain computing Oracle. Despite its
high potentials, o↵-chain scaling technology is still in its infancy with many technical
and economic challenges remaining unresolved.
To enable o↵-chain scaling for prime-time use, we propose Celer Network 1, a coher-
ent architecture that brings Internet scale to existing and future blockchains. Celer
Network consists of a carefully designed o↵-chain technology stack that achieves high
scalability and flexibility with strong security and privacy guarantees, and a game-
theoretical cryptoeconomic model that balances any new tradeo↵s.
1.1. Celer T echnology Stack
As a comprehensive full-stack platform that can be built upon existing or future
blockchains, Celer Network encompasses a cleanly layered architecture that decouples
sophisticated o↵-chain platform into hierarchical modules. This architecture greatly
simplifies the system design, development, and maintenance, so that each individual
component can easily evolve and adapt to changes.
A well-designed layered architecture should have open interfaces that enable and
encourage di↵erent implementation on each layer as long as they support the same
cross-layer interfaces. Each layer only needs to focus on achieving its own functionality.
Inspired by the successful layered design of the Internet, Celer Network adopts an o↵-
chain technology stack that can be built on di↵erent blockchains, named cStack,
which consists of the following layers in bottom-up order:
• cChannel: generalized state channel and sidechain suite.
• cRoute: provably optimal value transfer routing.
• cOS: development framework and runtime for o↵-chain enabled applications
1celer:s w i f t ,f a s ti nL a t i n ,t h ec for the speed of light
5
Figure 1. Celer Network layered architecture.
Celer architecture provides innovative solutions on all layers. Below we highlight
the technical challenges and features of cChannel, cRoute, and cOS.
cChannel. This is the bottom layer of Celer Network that interacts with di↵erent
underlying blockchains and provides the upper layer with a common abstraction of up-
to-date states and bounded-time finality. cChannel uses state channel and sidechain
techniques, which are both cornerstones of o↵-chain scaling platforms.
A state channel allows mutually distrustful parties to execute a program o↵-chain
and quickly settle on the latest agreed states, with their security and finality guaran-
teed by on-chain bond contracts. It was initially introduced by Lightning Network [9]
to support high-throughput o↵-chain Bitcoin micropayments. Since the introduction
of Lightning Network, there have been several research works that addressed di↵erent
problems in the context of payment channel networks, such as routing [10, 13, 16],
time lock optimization [5], and privacy [6]. However, o↵-chain network is still in its
early stage, facing a few major challenges in terms of modularity, flexibility, and cost-
eciency. cChannel meets current challenges by o↵ering a set of new features.
• Generic o↵-chain state transition. O↵-chain transactions can be arbitrary
state transitions with dependency DAG. This allows Celer Network to support
complex high-performance o↵-chain dApps such as gaming, online auction, in-
surance, prediction market and decentralized exchanges.
• Flexible and ecient value transfer. Multiple state channel and sidechain
constructions with di↵erent tradeo↵s on eciency and finality are provided to
support fast value transfer with generic condition dependency, minimal on-chain
interactions, and minimal fund lockup.
6
• Pure o↵-chain contract. Any contract that is not directly associated with
on-chain deposits does not need any on-chain operation or initialization unless
a dispute is triggered. Every pure o↵-chain contract or object has a uniquely
identifiable o↵-chain address, and only needs to be deployed on blockchains when
necessary with an on-chain address assigned by the built-in o↵-chain address
translator.
cRoute. Celer Network is a platform for highly scalable dApps, designed to support
high-throughput value transfer. O↵-chain value transfer is an essential requirement
of many o↵-chain applications. While Celer Network is capable of supporting dApps
beyond payment solutions, it also makes groundbreaking improvements to o↵-chain
payment routing as it directly determines how much and how fast value can be trans-
ferred within the ecosystem.
All of the existing o↵-chain payment routing proposals [3, 4, 10, 12, 13, 16] boil
down to conventional “shortest path routing” algorithms, which may achieve poor
performance in an o↵-chain payment network due to the fundamental di↵erences in
the link model. The link capacity of a computer network is stateless and stable (i.e.,
not a↵ected by past transmissions). However, the link capacity of an o↵-chain payment
network is stateful (i.e., determined by on-chain deposits and past payments), which
leads to a highly dynamic network where the topology and link states are constantly
changing. This makes conventional shortest path routing algorithms hard to converge,
and thus yields low throughput, long delay, and even outages.
To counter this fundamental challenge, Celer Network’s payment routing module,
cRoute, introduces Distributed Balanced Routing (DBR) , which routes payment
trac using distributed congestion gradients. We highlight a few unique properties of
DBR (details in § 3.2.2).
• Provably optimal throughput. We prove that for any global payment arrival
rate, if there exists a routing algorithm that can support the rate, then DBR can
achieve that. Our evaluation shows that DBR achieves 15x higher throughput
and 20x higher channel utilization ratio 2 compared to state-of-art solutions.
• T ransparent channel balancing. “Keeping the channels balanced” has been
our goal since the proposal of Lightning Network. However, existing attempts in
2the ratio between the amount of transferred fund in each time slot and the total deposits of all channels.
7
channel balancing comprise heuristics that require heavy on-chain or o↵-chain
coordination with poor guarantees. DBR embeds the channel balancing process
along with routing and constantly balances the network without requiring any
additional coordination.
• F ully decentralized. The DBR algorithm is a fully decentralized algorithm
where each node only needs to ”talk” to its neighbors in the state channel network
topology. DBR’s protocol also lowers messaging cost.
• F ailure resilience. The DBR algorithm is highly robust against failures: it
can quickly detect and adapt to unresponsive nodes, supporting the maximum
possible throughput over the remaining available nodes.
• Privacy preserving . The DBR algorithm can be seamlessly integrated with
onion routing [11] to preserve anonymity for sources/destinations. Due to its
multi-path nature, the DBR algorithm naturally preserves the privacy regarding
the amount of transferred value, without using any additional privacy-preserving
techniques (e.g., ZKSNARK).
cOS. An on-chain dApp is simply a frontend connecting to the blockchain. O↵-chain
dApps, though with great potentials for high scalability, are not as easy to build and
use as the traditional on-chain dApps. Celer Network introduces cOS which is a devel-
opment framework and runtime for everyone to easily develop, operate, and interact
with scalable o↵-chain dApps without being bogged down by the additional complex-
ities introduced by o↵-chain scaling. Celer Network allows developers to concentrate
on the application logic and create the best user experience, with the cOS dealing with
the heavy lifting including the following tasks.
• Figure out the dependency between arbitrary o↵-chain and on-chain states.
• Handle the tracking, storage, and dispute of o↵-chain states.
• Tolerate intermediate node failures transparently.
• Support multiple concurrent o↵-chain dApps.
• Compile a unified implementation to di↵erent on-chain and o↵-chain modules.
1.2. Celer Cryptoeconomics
Celer Network’s cryptoeconomic mechanism, cEconomy, is designed based on a fun-
damental principle: a good cryptoeconomic (token) model should provide additional
8
values and enable new game-theoretical dynamics that are otherwise impossible. While
gaining scalability, an o↵-chain platform is also making tradeo↵s on network liquidity
and state availability, and it will never take o↵ without a cryptoeconomic model that
can enable new dynamics to balance out these tradeo↵s.
New tradeo↵s. O↵-chain platform gains scalability by making the following tradeo↵s.
• Scalability vs. Liquidity . O↵-chain value transfer requires deposits to be
locked on-chain as network liquidity. This is especially challenging for potential
o↵-chain service providers, because a significant amount of liquidity is needed to
provide e↵ective o↵-chain services for global blockchain users, either as outgoing
deposits in state channels or fraud penalty bond in sidechains. However, owners
of a large number of crypto assets (whales) may not have the business interest or
technical capability to run an o↵-chain service infrastructure, while people who
have the technical capability of running a reliable and scalable o↵-chain service
often do not have e프로젝트로 돌아가기