Asynchronous approximate agreement

Publisher Website
Google Scholar
Abstract
This paper introduces an algorithm to solve the Approximate Agreement Problem in an asyn- chronous failure-by-omission (or crash-failure) system, and proves that the algorithm is optimal by consider- ing the power of the "adversary" scheduler to disrupt processors' views. We show that the adversary need not cause any crashes or omissions to achieve its purpose, and therefore no algorithm can do better than simply to operate round-by-round, as ours does. The result- ing understanding of the adversary should be applicable to other problems in asynchronous failure-by-omission or crash-failure systems.

This publication has 0 references indexed in Scilit: