Hardware Trojan Detection Method Based On Side Channel Analysys

Hardware Trojan Horses(HTHs or Trojans) are malicious design modifications intended to cause the design to function incorrectly. Globalization of the IC development industry has created new opportunities for rogue agents to compromise a design in such a way. Passive attacks on chips result in secret information leaking while active attacks cause IC malfunction and catastrophic system failures. Hardware Trojans do much more harm than software Trojans, determined by the physical characteristics of HTHs. HTHs have become one focus in the information security and hardware security, because of its huge harmfulness, though the history of researching on it is less than 10 years. Machine learning can summarizes the similarities and differences between different kinds of data, by studying the existing information. Then achieve the purpose of classifying and predicting the unknown data. As the core of artificial intelligence, machine learning is one of the forefront research field at present.Based on the Side Channel Analysis(SCA) technology, a set of hardware trojan detection platform is designed for RTL circuits on the basis of Hspice power consumption simulation and machine learning. Firstly, the hardware trojan detection and the implantation method of HTHs are studied. Then, a mini AES circuitis designed as a carrier circuit, in which three types of trojans, the foundation for experiment, are successfully implanted. Power acquisition platform is analysed based on Hspice and Monte Carlo. This platform can precisely simulate dynamic power consumption of integrated circuit and power fluctuations caus ed by different range of process deviation, which is more flexible than power acquisition platform on FPGA. On this basis, the process deviation of simulation is analysed and a compressing method of the mantissa data is put forward using PCA under the condition that an appropriate threshold is set. This method can compress the data dimension to less than 10% of the original one. It also can effectively reduce the influence of hardware Trojan detection technology caused by process deviation. Then the experimental data is transmitted into the artificial neural network to support vector for training and recognition. The result shows that the detection precision of PSO ANN is higher than the traditional ANN method and also the detection accuracy rate of grid optimization SVM is higher than that of GA SVM. The detection accuracy rate of grid optimization SVM is higher than that of PSO NN,but PSO NN is faster. Additionally, the result of recognition method based on SVM is more stable.