A method of noise optimization for Hardware Trojans detection based on BP neural network

The Hardware Trojans act as a new malicious attack for the underlying hardware, the research of method based on side-channel analysis become one of important contents for information security. But the method of Hardware Trojans detection based on side-channel analysis easily been affected by noise, and its detection model has low precision. This article uses the technology of BP neural network for Hardware Trojan detection, puts forward a method for the noise optimization by wavelet transform. We implant a hardware Trojan circuit occupied areas 0.54% in AES circuit, and make the realization of circuit and power analysis based on the FPGA platform. Experimental results show that the method of noise optimization by wavelet can eliminate the high frequency noise, improve the Hardware Trojan detection sensitivity based on neural network, and the detection sensitivity increased from 92.2% to 99.2%.