Research On Side-channel Analysis Based Hardwear Trojan Detection Technology

Information security has become a major developmental problem faced by the human society. Therefore, as an indispensable element of modern electronic equipment, the security of integrated circuit chips has been gaining more and more attention. Due to the increasing complexity and cost of the modern integrated circuit designing, its design and manufacturing could not be consistent with each other, further leading to new threats to integrated circuit products. Attackers could change the original design on the hardware level during the design or manufacturing process and embed Hardware Trojan in order to steal secret information, interrupt and destroy opponents’ central controlling systems, etc. Nowadays, from household appliances to national military, financial and transportation systems, all have been facing serious threats of Hardware Trojan. Thus, carrying out researches on Hardware Trojan detection technology is extremely important to chip security and information security.This thesis focuses on the side-channel analysis based Hardware Trojan detection technology, detecting both the known Trojan in AES benchmark and unknown Trojan in ARM9 via two different kinds of side-channel signal analysis methods. The results turn out to be satisfying and the findings are as follows:(1) Research on Hardware Trojan and its detection methods. Through analyzing the working principle and features of Hardware Trojan and summarizing both of the advantages and disadvantages of current Hardware Trojan detection technologies, this thesis focuses on the side-channel analysis based Hardware Trojan detection technology.(2) Physical design of circuits which will be detected. Finish back-end design of undetected circuits based on SMIC 130 nm CMOS technology. Analyze the effects on original circuit of Hardware Trojan during synthesis, place-and-route and formality phases, which is significant to success of Hardware Trojan detection.(3) Research on power analysis based detection technology. We use power as the side channel signal and detect 5 different Trojans embedded into AES cryptographic circuit by considering noise introduced by process variation ±10%. Fail of only the smallest Trojan which size is 0.01%, so to speak success of the Trojan which size is 0.12% and fully success of the other 3 Trojans. These results provide the effectiveness of our power analysis based detection technology.(4) Research on multiple-parameter analysis based detection technology. We use transient current and maximum operating frequency as the side channel signal and in the same process noise situation. Fail of only the smallest Trojan which size is 0.01% and fully success of the other 4 Trojans. Compared with power analysis, the effect of multiple-parameter analysis based detection technology has been highly improved ranging from 14.44% to 154.52%.