Split manufacturing of integrated circuits means to delegate the front-end-of-line (FEOL) and back-end-of-line (BEOL) parts to different foundries, in order to prevent overproduction, intellectual property (IP) piracy, or targeted insertion of hardware Trojans. SplitAttack challenges the security promise of split manufacturing by a sophisticated deep neural network that can infer the missing BEOL connections with high accuracy. In paticular, it features following method for an efficient and effective connection prediction:
- SplitExtract which formulates various layout-level placement and routing hints,
- a neural network makes use of vector-based and image-based layout features simultaneously,
- a loss function that directly and effectively select the most probable BEOL connection among the relevant candidates without suffering from an imbalance between positive and negative samples,
- ...
More details are in the following papers:
- Haocheng Li, Satwik Patnaik, Abhrajit Sengupta, Haoyu Yang, Johann Knechtel, Bei Yu, Evangeline F.Y. Young, Ozgur Sinanoglu, "Attacking split manufacturing from a deep learning perspective", ACM/IEEE Design Automation Conference (DAC), Las Vegas, NV, USA, June 2-6, 2019.
- Haocheng Li, Satwik Patnaik, Mohammed Ashraf, Haoyu Yang, Johann Knechtel, Bei Yu, Ozgur Sinanoglu, Evangeline F.Y. Young, "Deep Learning Analysis for Split Manufactured Layouts with Routing Perturbation", IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (TCAD), 2020.
$ git clone https://github.com/cuhk-eda/split-attack