The Sleepy Model of Consensus
- Detailed Technology Description
- We offer an enhanced blockchain consensus protocol, The Sleepy Model of Consensus, that doesn’t require proof-of-work andresiliency requires only a majority (>50%) of online nodes being honest
- Others
-
Publications
- Rafael Pass, Elaine Shi, “The Sleepy Model of Consensus”,Asiacrypt 2017, pp 380-409. https://link.springer.com/chapter/10.1007/978-3-319-70697-9_14
- Elaine Shi, “The Sleepy Model of Consensus” Asiacrypt 2017. https://www.youtube.com/watch?v=ARLbs2hDou4
- Elaine Shi RWC 2017 https://www.youtube.com/watch?v=B5EY65YP5p0
- *Abstract
-
Consensusprotocols are at the core of distributed computing and provide a foundation formulti-party cryptographic protocols (i.e. blockchains). Unlike traditionalsystems based on either synchronous models where honest messages are deliveredin rounds, or asynchronous models where honest messages are subject toadversarial delays. We take a hybrid approach that is an improvement to bothmethods and can (survive/account for) offline nodes as well as delayed messages.
Our protocol, the‘Sleepy Model of Consensus’ takes anapproach of categorizing honest nodes as alert (online with a good connection)or sleepy (some type of communication failure or delay) and other nodes ascorrupt. In this Sleepy consensus model, security or consensus can be maintained aslong as the alert nodes outnumber the corrupt nodes.
Most existingdecentralized blockchain-style cryptocurrencies utilize a permissionlessconsensus protocol based on proof-of-work which is computationally andenergetically wasteful, our model uses a more classical permissioned settingwithout proof-of-work.
Our protocolproceeds much like blockchains based on Nakamoto’s protocol, except instead ofsolving computational puzzles, a node can write to or extend the blockchain onlyif it is elected leader. Election to leader is based upon a random oracle or apseudo random function applied and a common reference string. The leader thencan extend the chain by including the previous blocks hash, the nodesidentifier in the form of a public key, the current timestamp, and thetransactions to be confirmed. Subsequent extension of the chain is performed bythe new leader which is determined based again on the timestamp, nodes publickey and the oracle.
This protocoltakes a ‘state machine approach’ for fault tolerance and to facilitatecooperation since there is no centralized control. This robust protocol accountsfor and determines time offsets or delays as a result of weak synchronizationof the nodes clocks. As part of the sleepy model of consensus, foreknowledge ofthe number of nodes is accomplished through a public key infrastructure wherenodes public keys are registered at the beginning of the protocol execution,prior to the determination of the leader.
Potential Applications
- Consortium Blockchain
- Proof-of-Stake consensus
- Cryptocurrencies
- Smart Contracts
Advantages
- Eliminates the need for proof-of-work
- Robust protocol tolerates up to 50% corruption (1/2 lowerbound)
Keywords: Physical Sciences, Engineering, Computer Software, Applications, Optimization, Computer Science, Computer Engineering, Cryptography, Cryptocurrency, Virtual Currency, Bitcoin, Blockchain, Algorithms, Network Security, Data Security
- *Licensing
- Martin Teschl, Technology Commercialization & Liaison Officermt439@cornell.edu(607) 254-4454
- Country/Region
- USA
