|Microcontrollers are extensively utilized across a diverse range of applications. However, with the escalating usage of these devices, the risk to their security and the valuable data they process correspondingly intensifies. These devices could potentially be susceptible to various security threats, with side channel leakage standing out as a notable concern. Among the numerous types of side-channel leakages, photon emissions from active devices emerge as a potentially significant concern. These emissions, a characteristic of all semiconductor devices including microcontrollers, occur during their operation. Depending on the operating point and the internal state of the chip, these emissions can reflect the device’s internal operations. Therefore, a malicious individual could potentially exploit these emissions to gain insights into the computations being performed within the device.
This dissertation delves into the investigation of photon emissions from the SRAM blocks of two distinct microcontrollers, utilizing a cost-effective setup. The aim is to extract information from these emissions, analyzing them as potential side-channel leakage points.
In the first segment of the study, a PIC microcontroller variant is investigated. The quiescent photon emissions from the SRAM are examined. A correlation attack was successfully executed on these emissions, which led to the recovery of the AES encryption key. Furthermore, differential analysis was used to examine the location of SRAM bits. The combination of this information with the application of an image processing method, namely the Structural Similarity Index (SSIM), assisted in revealing the content of SRAM cells from photon emission images.
The second segment of this study, for the first time, emphasizes on a RISC-V chip, examining the photon emissions of the SRAM during continuous reading. Probing the photon emissions from the row and column detectors led to the identification of a target word location, which is capable of revealing the AES key. Also, the content of target row was retrieved through the photon emissions originating from the drivers and the SRAM cells themselves. Additionally, the SSIM technique was utilized to determine the address of a targeted word in RISC-V photon emissions which cannot be analyzed through visual inspection.
The insights gained from this research contribute to a deeper understanding of side-channel leakage via photon emissions and demonstrate its potential potency in extracting critical information from digital devices. Moreover, this information significantly contributes to the development of innovative security measures, an aspect becoming increasingly crucial in our progressively digitized world.