A power analysis based approach to detect Trojan circuits

Abstract

Because of globalization of the semiconductor industry, the IC fabrication is increasingly outsourced. This poses a significant risk for integrated circuits (ICs) used for security critical applications. Attackers can maliciously alter the ICs during fabrication in untrusted foundries. In the case of ICs bought externally, they may have hidden functions that users would never know. These malicious alterations and hidden functions are also referred to as “Hardware Trojan”. It is extremely difficult to discover such Trojan circuits using conventional testing strategies. In this paper, we propose a nondestructive, power analysis based Trojan detection approach which is able to detect Trojan circuits in the presence of large noise. The approach is validated using 90nm FPGA (Xilinx Spartan-3E) chips. Experimental results with a 64-bit Data Encryption Standard (DES) cipher circuit show that Trojans which are 2 orders of magnitude smaller than the DES circuit can be detected by using statistic signal processing techniques.