Tendermint: Consensus without Mining
Jae Kwon
yk239@cornell.edu
Draft v.0.6 (outdated)
This is from 2014.
The current spec is on github.com/tendermint/tendermint/wiki
Abstract. Cryptocurrencies such as Bitcoin enable users to submit
payment transactions without going through a centralized trusted or-
ganization. Bitcoin relies on proof-of-work mining to secure consensus
which is problematic; mining requires a massive expenditure of energy,
confirmation of transactions is slow, and security is difficult to quantify.
We propose a solution to the blockchain consensus problem that does
not require mining by adapting an existing solution to the Byzantine
Generals Problem.
1. Introduction
Cryptocurrencies have come into the spotlight since the introduction of Bit-
coin [1]. The objective of cryptocurrency protocols such as Bitcoin is to maintain a
live decentralized transaction ledger while defending against double-spend attacks
from malicious Byzantine actors that deviate from the protocol. Consensus of the
Bitcoin transaction ledger is secured by a network of miners who compete for re-
wards in the blockchain. This mining, or proof-of-work, comes with a hefty cost.
At today’s Bitcoin prices and reward schedule, miners are rewarded on the order of
$1,500,000 a day to secure the blockchain – and a significant portion of that money
is spent on electricity. Proof-of-work based consensus protocols are also slow, re-
quiring up to an hour to reasonably confirm a payment to prevent double-spending.
Alternative protocols have been proposed by the cryptocurrency commu-
nity to replace proof-of-work mining [2, 3]. Some of these protocols suffer from the
nothing at stake problem; participants have nothing to lose by contributing to mul-
tiple blockchain forks, so consensus on a single blockchain is not guaranteed. Other
protocols require participant to accurately gauge the trustworthiness of other key
participants, which makes analysis of security difficult to quantify.
Our contribution is a novel consensus protocol that requires no proof-of-work
mining and has a high level of protection against double-spend attacks. We make
a weak assumption about the participant’s abilities to keep time, and we assume
partial synchrony of the network. Our algorithm is based on a modified version of
the DLS protocol [4], and is resilient up to
1
/3 of Byzantine participants.
1
剩余10页未读，继续阅读
评论0