automated counterfeit ic physical defect characterization
DESCRIPTION
Automated Counterfeit IC Physical Defect Characterization Team 176: Wesley Stevens, Dan Guerrera , Ryan Nesbit Advisors: Mohammad Tehranipoor , Domenic Forte ECE Department, University of Connecticut , { wesley.stevens , daniel.guerrera , ryan.nesbit }@ uconn.edu - PowerPoint PPT PresentationTRANSCRIPT
TEMPLATE DESIGN © 2008
www.PosterPresentations.comTEMPLATE DESIGN @ 2013 by Jifeng ChenCenter for Hardware Assurance, Security, and EngineeringUniversity of Connecticut
Automated Counterfeit IC Physical Defect CharacterizationTeam 176: Wesley Stevens, Dan Guerrera, Ryan Nesbit
Advisors: Mohammad Tehranipoor, Domenic ForteECE Department, University of Connecticut, {wesley.stevens, daniel.guerrera, ryan.nesbit}@uconn.edu
{tehrani, forte}@engr.uconn.edu
Motivation
Create an automated, user friendly program for identifying physical defects of ICs
Accept wide range of image inputs from various locations
Process different images with specific algorithms Compile and display comprehensive results
Objective and Solution
Increasing number of counterfeit integrated circuits (ICs) Counterfeit ICs can cause catastrophic failure of systems Current physical defect tests are destructive, time
consuming An expert is required both for performing tests and
analysis of results
General SpecificationsLanguage: MATLABAnalysis Types: Single, GoldenImage Types: Surface, Pin, TextImage Magnification: 20x – 100xIdeal Image Resolution: 1000 by 1000 pixelsOutput: Current Algorithm, Identified Defects, Summary
• Counterfeit determination is based on identifying defects or abnormalities with the IC• Physical defects can be categorized by the component
or location at which they occur• Imaging techniques provide data that can be used to
identify defects and determine IC authenticityDefects detected include: Pin: dents, contamination, color variations, misaligned Surface: scratches, color variation, improper textures,
package damage Text: markings, ghost markings
Surface Analysis Example Images
Feature Matching and Alignment
Pin Analysis
About the Authors
Expand Defect Coverage Improve Algorithm Robustness Expand Group Comparison Analysis Create Graphical User Interface Modify User Results
www.chase.uconn.edu
Transformation:
Algorithm Results:
Original Image:
Difference: 0.1103
Future Work
Wesley Stevens (EE/CE), Dan Guerrera (CE), and Ryan Nesbit (EE) are full time undergraduate students at the University of Connecticut.
Original Image:
Isolation of distinct objects:
Counting objects:
Scratch Analysis:• Counts results of all operations• Highlights areas with count greater than a given threshold
Statistical Averaging:• Cleans up excess blocks• Determines types of anomalies present in different blocks• Correlate types to various defects
Scratch Analysis:• Converts image to binary using threshold• Creates line structuring elements for comparison• Iterates through operations while varying parameters
Statistical Averaging:• Divides image into blocks based on size• Calculates Global and Local statistics• Compares each block to gathered statistics• Flags blocks outside of threshold
Approach and Methods
Object Isolation: •Uses differences in intensity values to find objects •Different structuring elements are used to find different objects•Algorithm iteratively grows these objects•The parameters of each structuring element are changed on each iteration •Given the type of structuring element the type of defect can be determined
•Algorithm will count and find the area of each object•This data is also used in determining what type of defects might exist •Certain checks exist to help filter out false positives