A novel physiological features-assisted architecture for rapidly distinguishing health problems from hardware Trojan attacks and errors in medical devices

Abstract

Malicious Hardware Trojans (HTs) that are inserted during chip manufacturing can corrupt data which if undetected may cause serious harm in medical devices. This paper presents a novel physiological features-assisted architecture to detect and distinguish attacks by ultra-small HTs from actual health problems in health monitoring applications. Our threat scenario considers attacks that pass undetected using other HT detection methods such as ones that use side-channel analysis and digital systems test. The key to our detection approach is to embed multiple signature generation and testing techniques, some of which are based on physiology, deep in the hardware and close to the origin of data generation. Our experimental results show that our proposed techniques are able to distinguish unhealthy physiology from functionality altering HT attacks anywhere inside a state-of-the-art medical chip including the chip's primary inputs with minimal performance and area overhead.