Matrix approach to perform dependent failure analysis in compliance with functional safety standards

Abstract

The main goal of Functional Safety is to implement accident avoidance, employee safety and machine safety systems. Emerging technologies like Artificial Intelligence (AI), Advanced Driver Assistance System (ADAS), autonomous driving and autonomous industries needs very high computing devices (like SoC) with functional safety implemented. This puts System on Chip (SoC) in critical path of functional safety due to its complex design and low process safety time. Every System on Chip Intellectual property (SoC IP) used in these market segments should be carefully analyzed and comply with automotive standard ISO26262 and industrial standard IEC61508, respectively. FMEDA is one such kind of analysis to identify design failure modes and its detection ability. Another analysis is dependent failure analysis (DFA) which explains SOC IP’s freedom from interference with respect to common cause failures or inactive/disable IPs associated with its design. Multiple authors proposed various ways to carry FMEDA analysis and automation. But best practices for dependent failure analysis and automation are still lacking. In this paper, we proposed a novel approach to perform dependent failure analysis using matrix method in compliance with industry standards. This novel method helps analyst to visualize interaction of IPs and enhances the quality of analysis. A new matrix method is used to execute DFA which gives systematic and attributable analysis.