GHOST
Papers
Original paper Secure High-Rate Transaction Processing in Bitcoin (2013)
Yonatan Sompolinsk and Aviv Zohar (The Hebrew University of Jerusalem)
FC'15
Speed-Security Tradeos in Blockchain Protocols
Aggelos Kiayias and Giorgos Panagiotakos (University of Edinburgh)
On Trees, Chains and Fast Transactions in the Blockchain
Aggelos Kiayias and Giorgos Panagiotakos (University of Edinburgh)
Latincrypt'17
Prove that GHOST satisfies liveness and persistence by a single, fresh block lemma
Lock-step
In fruitchain paper,
as the subsequent work by Kiayias and Panagiotakos shows, GHOST actually worsens “chain quality”
A Better Method to Analyze Blockchain Consistency
Lucianna Kiffer, Rajmohan Rajaraman and abhi shelat (Northeastern University)
CCS’18
See in Nakamoto consensus
Number of confirmations
Number of confirmation blocks for Bitcoin and GHOST consensus protocols on networks with delayed message delivery
CryBlock'18 Slide, Extended abstract
Lyudmila Kovalchuk, et al.
Prior work: Analysis of Spliting Attacks on Bitcoin and GHOST Consensus Protocols
Comparison of Block Expectation Time for Various Consensus Algorithms
Kaidalov D. S. (IOHK) et al.
November/2018, Radio Electronics, Computer Science, Control (Ukranian Journal)
Average block confirmation time for Bitcoin, GHOST, Ouroboros
Blancing, splitting
The Balance Attack Against Proof-Of-Work Blockchains: The R3 Testbed as an Example
Christopher Natoli, Vincent Gramoli (University of Sydney)
Analyze balancing attack (to keep forking in GHOST) by an adversary with the ability to divide the network into subgraphs with equal hash power
Calculate the probability that this attack makes safety failure for double spending oppotunity
Experiment on R3 consortium blockchain
Variants
LMD GHOST (Ethereum)
GHAST: Breaking Confirmation Delay Barrier in Nakamoto Consensus via Adaptive Weighted Blocks
Chenxing Li (Tsinghua University), Fan Long (University of Toronto), Guang Yang (Conflux Foundation)
Confirmation Policy (From Section 4.3 of the Conflux paper)
A block b is confirmed if for any ancestor block of b, the corresponding subtree total weight is heavier than all of the subtrees of its siblings by a margin.
With the parameter setting used in our experiments, this margin is about 20∼30
in normal scenarios for obtaining the same confidence as waiting six blocks in Bitcoin.
Even faster block-chains with DECOR protocol
NB: Ethereum doesn't use GHOST
Stackexchange: Ref, Ref, Ref
ethresear.ch Are Uncle Blocks (ommers) included in total chain difficulty / weight?
#Layer1