| Hardware Trojan problem has become an important issue, which must be considered in the information security and protection. With the fast development of the Integrated Circuit(IC), chips have been used in the daily life. Therefore, the security of the chips is important to the development of the national economic and politics, the construction of national defense and human’s information security. Thus, it’s significant to carry out the Hardware Trojan detection techniques based on the side channel analysis.The research emphasis of this thesis is the detecting techniques of Hardware Trojan based on the side channel analysis under different process variation. The processing method of the power consumption data is proposed to reduce process noise and data redundancy. This method can improve the SNR(Signal to Noise Ratio) of the Trojan circuits effectively. Besides, the method of improving the Trojan actication probability is also studied in-depth, which can assist the detection of side channel analysis. The main research contents of this thesis are as follows:1. The influence of different process variation on the circuits is simulated. The influence of different process variation on the power consumption and time-delay have been studied. By modifying the Gauss parameters of the process library, three kinds of process variation have been set up, and the corresponding current variations(CV) are ±2%, ±5% and ±7%, respectively. Among them, ±5% is the maximum variation of SMIC 130 nm. The average transient current analysis has been carried out on four circuits and corresponding Trojan circuits with the process variation simulation method based on the Monte Carlo simulation. When CV is ±7%, the smallest Trojan which can be detected is 2%; When CV is ±5%, the smallest Trojan which can be detected is 1.5%; When CV is ±2%, the smallest Trojan which can be detected is 0.8%, the detection efficiency is very low which means that the process variation has great interference on the Hardware Trojan detection.2. The research on the method of data processing of Hardware Trojan detection based on power consumption analysis. The model of Trojan detection has been established, which is suitable for signal feature extraction. The methods of data’s dimension reduction and noise reduction based on K-L transformation and Mahalanobis distance analysis have been studied in detail. The combination of the two methods can effectively improve the SNR of the Trojan circuits. The experimental results show that the degree of separation between the processed original circuits’ data and the processed Trojans’ circuit data has been improved significantly compared with the unprocessed power consumption, when CV is ±7% and ±5%, the smallest Trojan which can be detected is 0.6%; when CV is ±2%, the smallest Trojan which can be detected is 0.34%.3. The research on enhancing the activation probability of Hardware Trojan is carried out. By analyzing the influencing factors of Trojan activation probability, a new structure which can improve the Trojan’s activation probability has been inserted into the circuit, and the insertion algorithm of this structure has been improved to reduce the effect of this structure on the circuit’s area, and the maximum area optimization rate has been reached to 44.8%. The simulation and power consumption analysis can proved that this structure has increased all nodes’ turnover rate in circuits to threshold(thP) above with the condition of circuits’ area increasing less than 0.44%, increased the switching probability of the internal nodes of the Trojan circuits, and reduced the activation time of Trojan circuit and increased the activation probability of the Trojan circuits which make the Trojan circuits’ power consumption rate increases an average of 0.39% contribution to the total power consumption. Thus, this structure can assist the Hardware Trojan detection based on the side channel analysis effectively. |
PDF Full Text Request