Digital Forensic of 3D Printer

  • Sarah Alyammahi

Student thesis: Master's Thesis

Abstract

Nowadays 3D printing technology is rapidly replacing traditional manufacturing methods. Consequently, malicious users are also leveraging the new technology in producing unlawful or counterfeit items. Therefore, there is a need to trace the origin of unlawful or counterfeit 3D printed objects via digital forensics. The traditional countermeasures such as 'PrinTracker' rely heavily on textural parameters for forensics; other methods like watermarking that rely on monitored manufacturing processes are impractical when tracing 3D items to the source printer. The researcher argues that 3D printers have unique fingerprints because of hardware imperfections that lead to non-uniform dimensions and weight, density, of the 3D printed physical object despite using the same Computer Aided Design (CAD) model. A CAD model of a knife was printed from three different 3D Ultimaker printers. The knife specimens were printed using polylactic acid (PLA) material. Each of the three printers produced three copies of the 3D printouts. Optimal control of the 3D printing machines ensured the reliability of the experimental result of the 3D printed objects. Weight measurements were taken to verify variation in density while width, length, thickness, circular and curve diameters to check variation in dimensions of the 3D printouts if existed among the 3D printers. The ANOVA and post hoc test, Student-Newman-Keuls, were used when checking if the parametric means of weights and dimensions were statistically similar. The experimental result reveals that there is a significant difference in weights and dimensions of objects from the different 3D printers. Eight fingerprints, signature, based on weight and dimensions are experimentally presented and categorized based on the statistical test results. The repeated variation between the two Ultimaker 3 Extended and Ultimaker 2+ show unique fingerprints associated with 3D printers. K-Nearest Neighbor mining method is suggested by the researcher as the best method for future criminal investigation involving the three 3D printers, since experimental database mining is the best way of producing a reasonable and unambiguous pattern in digital forensic of 3D printer.
Date of AwardJul 2019
Original languageAmerican English

Keywords

  • 3D printing
  • ANOVA
  • forensics
  • K-Nearest Neighbor Algorithm
  • PLA
  • Student-Newman-Keu

Cite this

'