Detection of Malicious Circuit in Hardware using Compressive Sensing Algorithm

Abstract

It has been observed very critically in the hardware that robust Trojans are inserted in outsourced products resulting in security vulnerabilities. It is more important to note that some areas are in need to be more reliable and trust worthy with respect to security. A perpetrator in any possible location could alter the circuit behavior by including an unintended module in the circuit which can be very hard or impossible to detect until triggered. In this paper, a self-referencing approach is proposed that measures the leakage power of the circuit at different time windows. This approach can efficiently and reliably detect the leakage power that the Trojan creates and also avoids the requirement of golden chip by mapping a chip’s leakage power at different time window. The proposed method is also formulated using compressive sensing algorithm for selecting Trojan revealing test vectors and compresses the chip output. The proposed method is validated on ISCAS’85 benchmark circuits, in which 82% reduction in the output dimensions for the leakage power measurement is achieved for a compression ratio of 0.75.