Unanticipated Hardware Trojan Detection Technique Based on Antenna Reflection Coefficient Features Outside of its Operation Frequency

The involvement of different entities in the life cycle of wireless electronics has increased the risk of adversary attacks, in particular, hardware Trojans (HTs). Wireless network hardware need to be compliant with a set of minimum security requirements to protect the data that these systems exchange and ensure the system reliability, however, HTs make them vulnerable. Hardware-based malicious attacks and defense mechanisms are continuously being analyzed to provide prevention and detection capabilities against them. Considering the unpredictability of HTs, there is a need for blind countermeasures that can detect and identify HTs without any previous knowledge of their characteristics: fingerprint-based methods. This paper presents a technique for the detection of unanticipated HTs based on antenna input reflection coefficient measurements (S
11
) over a wide range of frequencies expanding far beyond the antenna operation frequency. This work includes the design, manufacturing and testing of printed circuit boards with a WiFi system-on-chip (ESP8285), a meandered inverted-F antenna, and a HT that shorts the antenna to disrupt the communication link. The effects of the insertion of the HT in both operational and non-operational modes are successfully used to detect its presence without anticipation of its characteristics using data similarity and distance measures (Pearson’s coefficient, Euclidean and Manhattan distances) that can be extrapolated to machine learning algorithms for large scale analyses.