Statistical Variability Analysis of SilGeo

Abstract

The electronics manufacturing and supply industry is rife with the unchecked circulation of counterfeit and tampered components. Although developments have been made to counter this issue, the methods often involve visual analysis, X-ray imaging, destructive testing or extensive functional verification of individual components. Applying one or more of these solutions to verify electronic components requires a significant investment in time and capital.

In this thesis, we present a statistical evaluation of SilGeo, a non-intrusive counterfeit electronics detection technology presented by Moreno et al. The primary purpose of this evaluation is to progress towards transforming SilGeo from a successful research result into a standardised verification method in industry applications.

Considering SilGeo as a measurement system, we focus on quantifying the variability due to the hardware assembly and determining repeatability and reproducibility. We use Design of Experiments concepts and hardware domain knowledge to identify main effects and generate assembly configurations to be tested. Accordingly, we apply the statistical technique analysis of variance (ANOVA) to obtain the variability results with individual factor contributions.