A Review Paper on Machine Learning Based Trojan Detection in the IoT Chips

Abstract

AbstractRecently, Internet of Things (IoTs) are in wide use and the IoTs are referred as growing network of devices which establish an internet connectivity and communication among the devices. The IoT uses billions of data as it connects billions of devices to the internet. The IoT has different layers such as sensing layer, physical layer and application layer, each layer undergoes various method of security. However, the basic fundamental of IoTs are hardware, thus there is a more concern to secure the hardware from the adversary attack as the Integrated Circuits (ICs) manufacturing flow are globalized. The adversary attacks may cause a malicious functions like change in the functionality of the circuit, leakage of data, reduce reliability, modification of parameters etc. hence, there is a need for securing the hardware this is known as Hardware Security. In recent times, many researchers as come up with various methodologies to secure the Hardware by detecting the presence of Trojan in the Hardware. One such methodology is the Machine Learning based Hardware Trojan detection, where various classifiers are trained to detect the small Trojans present in the Hardware circuit. Thus, in this article we are going to highlight the important aspects of Machine Learning Classifier over the Hardware Security problems. Here we are going to compare various Machine Learning Classifiers used so far in the Hardware Trojan Detection problem.