Definition and Empirical Evaluation of Voters for Redundant Smart Sensor Systems

Voting algorithms are used to arbitrate between the results of redundant modules in fault-tolerant systems. Smart sensors are used for FDI (Fault Detection and Isolation) purposes by means of their built in intelligence. Integration of fault masking and FDI strategies is necessary in the construction of ultra-available/safe systems with on-line fault detection capability. This article introduces a range of novel software voting algorithms which adjudicate between the results of redundant smart sensors in a Triple Modular Redundant, TMR, system. Techniques to integrate replicated smart sensors and fault masking approach are discussed and a classification of hybrid voters is provided. Voters are classified into four groups: Independent-diagnostic safety-optimised voters, Integrateddiagnostic safety-optimised voters, Independent-diagnostic availability-optimised voters and Integrated-diagnostic availability-optimised voters. The properties of each category are explained and sample versions of each class as well as their possible application areas are discussed. This study is the first attempt for integration voting algorithms with fault diagnosis devices. Experimental results are presented in this paper.