New Partitioning Approach for Hardware Trojan Detection Using Side-Channel Measurements

Hardware Trojans have emerged as a security threat to many critical systems. In particular, malicious hardware components can be inserted at the foundry for implementing hidden backdoors to leak secret information. In this paper, we present a new method to partition the circuit under test into blocks in order to obtain different side-channel signatures per chip. Each signature indicates which block is off or on in terms of the dynamic power switching activity. As a result, there are different co-existing decisions to more precisely detect the Trojan instead of one decision resulting from one side-channel signature. Moreover, this method detects in which block the Trojan exists. AES was used as an example to be divided into blocks. Sakura-G was used as an implementation target. The obtained results give four decisions to enhance Trojan existence and position. This paper also presents a methodology for Trojan detection using a cryptographic protocol to secure the detection process.