Implementing Oracles in Asynchronous Systems Prone to Process Crash Failures

Abstract

The notion of a failure detector has been introduced in Section 3.3. Considering a communication or agreement abstraction which is impossible to solve in the basic model CAMP_n,t$[\emptyset]$, an appropriate failure detector provides the processes with additional computability power, which allows this communication or agreement abstraction to be implemented in the corresponding enriched model. Various failure detectors have been presented and used in previous chapters (in Chap. 3 to implement URBbroadcast for any value of t despite fair channels, in Chap. 7 to implement a read/write register for any value of t, and in Chap. 17 to implement consensus despite asynchrony and process crashes).