Integration and Comparison of FDI and Fault Masking Features in Embedded Systems

Abstract

This paper addresses the benefits of integrating fault masking (FM) and fault detection and identification (FDI) methods to an embedded control system. It is shown that using a “smart” sensor in a Triple Modular Redundancy (TMR) configuration improves the system performance. Within the proposed hybrid system, “smart” sensors are used as redundant modules of the TMR configuration. A novel quality measure, which indicates the confidence level of a “smart” sensor output is used by a purpose-built voting algorithm to produce outputs. The proposed voting algorithm is a modification of weighted average voter, which produces results based on both the sensor output and the associated quality measure. This algorithm performs better than the conventional weighted average voter within a simple TMR configuration for low error ratios (<16). The main reason for this improvement is the capability of hybrid system to mask similar faults. Therefore, the proposed hybrid system produces more reliable results than the fault masking approach.