Localization of Electromagnetic Sources in Complex Processors

Abstract

Recent advances in computer hardware security research have shown that electronic devices are vulnerable to electromagnetic (EM) side-channel leakage. An attacker can steal sensitive information from an electronic device that is not connected to the network just by measuring the electromagnetic emanations produced by the device. The feasibility of a successful EM side-channel attack has grown significantly in recent years with the reduction in cost of electronic test equipment and the proliferation of electronic devices. Smartcards, cellphones, and laptops can be susceptible to side-channel attacks, and with the trend towards internet-of-things (IoT) and millions of connected devices, the side-channel will become of increasing security importance. While most of the current literature in electromagnetic side-channels focuses on demonstrating novel attack methods on various devices, relatively little research focuses on characterizing the EM emanations. Research presented in this thesis aims to develop a metric to measure the side-channel energy per instruction available to an attacker. The importance of this is to understand the effect of instruction level events (i.e. ADD, LOAD, NOP) on the electromagnetic emanations from a cellphone. Understanding the correlation between the instruction run and the corresponding electromagnetic footprint can help to understand which operations cause the greatest amount of electromagnetic leakage as well as provide a basis of measurement. In addition, a method of localization is proposed to identify the source of the EM emanations. A magnetic dipole equivalent source model is proposed and measurement results are compared with simulation results of the model. These results aim to provide a deeper insight into the nature of electromagnetic side-channels as well as provide hardware and software designers with the information needed to combat unnecessary EM emanations.