Impossibility of distributed consensus with one faulty process
Journal of the ACM (JACM)
Distributed systems (3rd ed.): concepts and design
Distributed systems (3rd ed.): concepts and design
Practical byzantine fault tolerance and proactive recovery
ACM Transactions on Computer Systems (TOCS)
Random Key Predistribution Schemes for Sensor Networks
SP '03 Proceedings of the 2003 IEEE Symposium on Security and Privacy
A Byzantine Fault Tolerant Distributed Commit Protocol
DASC '07 Proceedings of the Third IEEE International Symposium on Dependable, Autonomic and Secure Computing
| Hi-index | 0.00 |
Consensus algorithms that, essentially, endeavor agreement or commit on a particular transaction, are preeminent building blocks of distributed systems. It poses more severe threats, in case, the distributed network possesses some arbitrary behaving (malicious) nodes. The proposed article adds up a non-faulty agreement decision to the requesting client nodes from the coordinator replicas. The work is divided into two-phases, namely, a fault-free and fair cluster formation which employs authenticated key management scheme and secondly, an authenticated agreement among the cluster heads resulting in secure and correct outcome of a transaction. We assume a two-layer hierarchy with different clusters of replicas in one layer associated with their cluster heads on another layer. Multiple levels of encryption are incorporated by means of two keys: (1) a unique pair-wise key between the processes, and (2) a communication key that provides more authenticity and enables a secure communication among the processes. The necessary correctness proof has also been presented. The protocol is robust and exhibits better efficiency for long-lived systems.