Integrating selected fault masking and self-diagnosis mechanisms

Fault Detection and Isolation (FDI) using self-diagnosing elements is used in distributed systems to facilitate system monitoring and maintenance. Voting algorithms are used to increase reliability by masking faults. This research addresses the integration of self-diagnosing elements and voting algorithms in a Triple Modular Redundant (TMR) configuration. A self-diagnosing element model is presented and the problem of masking element faults is considered. A range of integrated (or hybrid) voters which handle diagnosis information are introduced. Experimental results show that the appropriate use of diagnosis in a fault masking system enables the voter to select more correct results than voting on element results values alone. Three categories of architecture are proposed which have ability to mask certain self-diagnosing element common error modes. This approach results in a modest improvement in the safety and reliability levels of the application. The hybrid voters proposed capture certain advantages of both fault masking and fault detection and isolation features