Research On Key Technology Of Hidden Hardware Trojan Multidimensional Diagnosis

Since the 21 st century,Integrated Circuit(IC)has gradually become an important part of modern society.No matter in military equipment or civilian products,chips are widely used and indispensable.However,in the era of globalization,there are often many untrusted third parties involved in the manufacturing process of large-scale chips.A growing number of hackers are implanting malicious Hardware Trojan into chips to alter their original function,degrade performance or steal internal important information.As a result,the chip is often not controlled by the design company,and the chip security has gradually become the primary problem of the chip.In this thesis,the Hardware Trojan of large-scale chip is studied,the basic concept and structure of Hardware Trojan are described and classified according to the characteristics of Hardware Trojan.Then,this thesis compares and analyzes several existing Hardware Trojan detection technologies,and summarizes their advantages and disadvantages.It is found that due to the rapid development of chip manufacturing technology,the area of IC is smaller,the design is more complex,and the scale is larger,which makes it difficult to extract the complete netlist of the chip with existing reverse engineering,and using the current Hardware Trojan detection technology to detect the gate-level netlist of largescale chips,it is difficult to give the detection results in a short time.Aiming at these two problems,this thesis proposes two kinds of hardware backdoor detection technology to solve the problem of unable to detect incomplete netlists and shorten the time of largescale chip Hardware Trojan detection.The specific contents are as follows:(1)The Hardware Trojan detection technology for incomplete gate-level netlist is proposed.Aiming at the problem that incomplete gate-level netlists cannot be detected,the thesis presents a Hardware Trojan detection technology for incomplete gate-level netlist.Firstly,the technology extracts the incomplete information in the gate-level netlist of the chip,and uses the circuit compensation algorithm to repair the incomplete circuit information.Secondly the technology use the SCOAP algorithm to calculate the node testability values in the circuit information,and input the node testability data into the K-means neural network for clustering,and finally analyze the clustering results to determine whether the chip netlist contains Hardware Trojan.This technology is used to test and analyze the Trust-Hub test set,which is the most recognized in the industry.The results show that this technology can detect the Trojan-free netlist in Trust-Hub test set without false alarm phenomenon.For the gate-level netlist files whose missing degree is less than 5%,the Hardware Trojan detection rate is up to 95%? For the gate-level netlist files whose missing degree is less than 10%,the Hardware Trojan detection rate is not less than 90%?(2)The Hardware Trojan detection technology based on large-scale chip gate-level netlist is proposed.Aiming at the problem that the current Hardware Trojan detection technology cannot provide detection results in a short time,this thesis proposes a Hardware Trojan detection technology based on large-scale chip netlist.Firstly,the technology extracts the node information in the gate-level netlist,calculates the circuit depth of each node,and divides the chip circuit into several combinational circuit modules according to the flip-flops in the circuit.Secondly according to the circuit depth of the node,set the initial value of the node testability values,and use the SCOAP algorithm to calculate the node testability values.Finally,the K-means clustering algorithm is used to determine whether there is Hardware Trojan in the netlist.Finally,the Trust-Hub test set is taken as the test sample to compare and analyze the detection time and detection effect of the existing Hardware Trojan detection technology COTD(Reference-Free Hardware Trojan Detection and Recovery Based on Controllability and Observability in Gate-Level Netlist)and this technology.The experimental results show that if the same gate-level netlist file is detected,the Hardware Trojan detection technology based on large-scale chip gate-level netlist only needs one quarter of the detection time of COTD,and the time cost is greatly reduced.