Characteristic Reassignment for Hardware Trojan Detection

With the current business model and increasing complexity of hardware designs, third-party Intellectual Properties (IPs) are prevalently incorporated into first-party designs. However, the use of third-party IPs increases security concerns related to hardware Trojans inserted by attackers. A core threat posed by Hardware Trojans is the difficulty in detecting such malicious insertions/alternations in order to prevent the damage. This thesis work provides major improvements on a soft IP analysis methodology and tool known as the Structural Checking tool, which analyzes Register-Transfer Level (RTL) soft IPs for determining their functionalities and screening for hardware Trojans. This is done by breaking down primary ports and internal signals into assigned assets that are spread out into six characteristics. Using characteristics based on the external primary ports and the internal signals connected to them, reassignment of assets can be used to match against entries using coarse-grained-to-coarse-grained matching against a subset of known-IPs to classify an unknown soft IP. After determining the soft IP’s functionality, asset reassignment occurs within the Golden Reference Library (GRL), a library consisting of known Trojan-free and Trojan-infested entries. A fine-grained-to-fine-grained asset reassignment is used against the GRL to contain the most up-to-date assets based on the unknown soft IP, where the matching process is used to determine if the soft IP is Trojan-free or Trojan-infested. With the increasing size of the GRL, the need to decrease computational resources while also maintaining high accuracy between unknown soft IPs and GRL entries is vital