Stealth assessment of hardware Trojans in a microcontroller

Abstract

Many experimental hardware Trojans from the literature explore the potential threat vectors, but do not address the stealthiness of the malicious hardware. If a Trojan requires a large amount of area or power, then it can be easier to detect. Instead, a more focused attack can potentially avoid detection. This paper explores the cost in both area and power consumption of several small, focused attacks on an Intel 8051 microcontroller implemented with a standard cell library. The resulting cost in total area varied from a 0.4% increase in the design, down to a 0.150% increase in the design. Dynamic and leakage power showed similar results.