A failure detector is called l-accurate if in each run at least l correct processes are unsuspected. (The property implies a bound of N - l on the resiliency; weak accuracy is I -accuracy.) Adapt...

A failure detector is called l-accurate if in each run at least l correct processes are unsuspected. (The property implies a bound of N - l on the resiliency; weak accuracy is I -accuracy.)

Adapt Algorithm 16.1 so as to complete in N - l + 1 rounds.

Prove that Dijkstra's token ring reaches a legitimate configuration in O(N2 ) steps. Shorten the analysis by giving a single norm function, quadratically bounded in N, that decreases with every step of the algorithm.