Scalable segmentation-based malicious circuitry detection and diagnosis

Hardware Trojans (HTs) pose a significant threat to the modern and pending integrated circuit (IC). Several approaches have been proposed to detect HTs, but they are either incapable of detecting HTs under the presence of process variation (PV) or unable to handle very large circuits in the modern IC industry. We develop a scalable HT detection and diagnosis scheme by using segmentation techniques and gate level characterization (GLC). In order to address the scalability issue, we propose a segmentation method which divides the large circuit into small sub-circuits by using input vector control. We propose a segment selection model in terms of properties of segments and their effects on GLC accuracy. The model parameters are calibrated by sampled data from the GLC process. Based on the selected segments we are able to detect and diagnose HTs correctly by tracing gate level leakage power. We evaluate our approach on several ISCAS85/ISCAS89/ITC99 benchmarks. The simulation results show that our approach is capable of detecting and diagnosing HTs accurately on large circuits.