Supervised Feature Extraction and Synthesize of Integrated Circuits Micrographs for Physical Assurance
Supervised Feature Extraction and Synthesize of Integrated Circuits Micrographs for Physical Assurance
Embedded Files
The target of the task is to analyze SEM images of 28nm to detect and classify logical cells to find out alteration of any kind
Introduction
Introduction
The goal of the project is to build up necessary components to facilitate the design of a real-time trojan detection system. The task has been initiated on the SEM images of 28nm Node Technology and will be extended on other nodes (e.g., 14nm) in future. The project focuses on detecting anomaly inserted by third party foundries. The real-time system constitutes of a cell extraction, cell identification part and a decision analysis part. The task involves extracting logical cell images from SEM image and use those to generate diversified synthetic cell images for different illumination condition. Using these real and synthetic images, the cell recognition unit is trained and prepared to deploy in the real-time system. The full outline, from SEM image capture to trojan detection will be presented as a separate publication in future. That's why we are limiting here the discussion up-to the currently published task [1].
System Flow
System Flow
SEM images are captured from the logical region of IC's. Images are pre-processed and logical cells are extracted from the images (R.O.I extraction part in the image below). These images will later be classified, and the output of the classifier will be matched against the entries of DEF file.
The number of cell images we acquire from SEM images are not sufficient to train a robust classifier. So, using the extracted cell images, we generate synthetic cell images using generative adversarial network (GAN). These synthetic and extracted cell images constitute the dataset for future training (see image below). The process is depicted in the figure below. This work has been accepted in the IEEE International Symposium on the Physical and Failure Analysis of Integrated Circuits (IPFA), 2021.
Our future task is to train a robust logical cell recognition unit. During inference time, we will extract cell images from particular position of an SEM image, pass it through classifier and match the classifier output with the DEF file entry of the corresponding position. Upon matching the output with the DEF file entry, probable trojan presence will be identified.
Related Publications
Related Publications
[1] M. M. Al Hasan, N. Vashistha, S. Taheri, M. Tehranipoor, and N. Asadi, “Generative Adversarial Network for Integrated Circuits Physical Assurance Using Scanning Electron Microscopy” 2021 IEEE International Symposium on the Physical and Failure Analysis of Integrated Circuits (IPFA), 2021
Page updated
Google Sites
Report abuse