...[8].

The protocol presented here is a simplified version of that given in the book of Nancy Lynch [27]. There it is claimed that the protocol needs the assumption f<n/3 but this is incorrect. As we shall see, f<n/2 suffices.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...

This is the title of the original paper by Chandra and Toueg which introduced this approach [11]. Unreliable failure detectors are now an active area of research.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...theorem.

The nice proof used here was communicated to me by Alexander Russell.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.