Adaptive Fault Diagnosis using Self-Referential Reasoning

TL;DR

This paper introduces a self-referential reasoning approach for fault diagnosis in remote processors, effectively identifying reliable units even when less than half are truthful, by treating truth-tellers and liars as reliable.

Contribution

It presents a novel self-referential reasoning method that improves fault diagnosis in complex scenarios with mixed processor behaviors.

Findings

Effective identification of reliable processors with fewer than half being truthful

Utilizes self-referential questions to distinguish processor types

Builds on earlier techniques to handle more challenging fault scenarios

Abstract

The problem is to determine which processors are reliable in a remote location by asking "Yes or No" questions. The processors are of three types: those that always tell the truth, those that always lie, and those the sometimes tell the truth and sometimes lie. Using self-referential reasoning, along with earlier techniques, we can regard both the truth-tellers and liars as reliable and thus the tackle situations when fewer than half the processors are truth-tellers.