Securing IoT Chips from Hardware Trojan using Machine Learning Classifiers

Recent trend shows that there is a widespread of Internet of Things (IoTs). IoTs are a network of devices that establish interconnectivity between them and communicate with each other. These IoTs had impacted every person in their daily lives through various applications such as smart homes, smart vehicles, smart medical, etc. The growing applications show the involvement of hardware devices and these devices are prone to attacks from the adversary. Hence, there is a need for securing the devices which in turn secure the IoTs. The attacks from adversaries are more in recent days to access secret information, perform Denial-of-service, performance degradation, etc. these attacks are performed as per the intention of the adversary. These types of attacks through hardware are called Hardware Trojan (HT) attacks. Hence, there is a requirement of security in a hardware device which is achieved by applying the non-destructive machine learning classifier method. This proposed methodology detects the HT present in a circuit by classifying the different parametric features of the circuit under test by differentiating the unknown and known netlists and detects the particular net as a ‘Trojan net’ or ‘normal net’, with the achievement of 94.4% of accuracy and 0.9 of f-measure.