Byzantine-Tolerant Causal Ordering for Unicasts, Multicasts, and Broadcasts

Byzantine fault-tolerant causal ordering of messages is useful to many applications. Causal ordering requires a property that we term strong safety, and liveness. In this paper, we use execution histories to prove that it is impossible to solve causal ordering - strong safety and liveness - in a det...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE transactions on parallel and distributed systems Ročník 35; číslo 5; s. 814 - 828
Hlavní autoři: Misra, Anshuman, Kshemkalyani, Ajay D.
Médium: Journal Article
Jazyk:angličtina
Vydáno: New York IEEE 01.05.2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Témata:
Algorithms
asynchronous system
broadcast
Broadcasting
Byzantine fault-tolerance
causal order
Cryptography
Fault tolerance
History
Messages
multicast
Multicast algorithms
Safety
Synchronization
Unicast
Upper bound
Upper bounds
ISSN:1045-9219, 1558-2183
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Byzantine fault-tolerant causal ordering of messages is useful to many applications. Causal ordering requires a property that we term strong safety, and liveness. In this paper, we use execution histories to prove that it is impossible to solve causal ordering - strong safety and liveness - in a deterministic manner for unicasts, multicasts, and broadcasts in an asynchronous system with one or more Byzantine processes. We also define a weaker version of strong safety termed weak safety. We prove that it is impossible to solve causal ordering - weak safety and liveness - in a deterministic manner for unicasts and multicasts, in an asynchronous system with one or more Byzantine processes. In view of these impossibility results, we propose the Sender-Inhibition algorithm and the Channel Sync algorithm to provide causal ordering - weak safety and liveness - of unicasts under the Byzantine failure model in synchronous systems, which have a known upper bound on message latency. The algorithms operate under the synchronous system model, but are inherently asynchronous and offer a high degree of concurrency as lock-step communication is not assumed. The two algorithms provide different trade-offs. We also indicate how the algorithms can be extended to multicasts.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1045-9219
1558-2183
DOI:10.1109/TPDS.2024.3368280