Fault injection attacks (FIA) are a class of active physical attacks, mostly used for malicious purposes such as extraction of cryptographic keys, privilege escalation, attacks on neural network implementations. There are many techniques that can be used to cause the faults in integrated circuits, many of them coming from the area of failure analysis. In this paper we tackle the topic of practicality of FIA. We analyze the most commonly used techniques that can be found in the literature, such as voltage/clock glitching, electromagnetic pulses, lasers, and Rowhammer attacks. To summarize, FIA can be mounted on most commonly used architectures from ARM, Intel, AMD, by utilizing injection devices that are often below the thousand dollar mark. Therefore, we believe these attacks can be considered practical in many scenarios, especially when the attacker can physically access the target device.
6 September 2021 in IEEE Transactions on Reliability
Neural networks have been shown to be vulnerable against fault injection attacks. These attacks change the physical behavior of the device during the computation, resulting in a change of value that is currently being computed. They can be realized by various techniques, ranging from clock/voltage glitching to application of lasers to rowhammer. Previous works have mostly explored fault attacks for output misclassification, thus affecting the reliability of neural networks. In this article, we investigate the possibility to reverse engineer neural networks with fault attacks. Sign bit flip fault attack enables the reverse engineering by changing the sign of intermediate values. We develop the first exact extraction method on deep-layer feature extractor networks that provably allows the recovery of proprietary model parameters. Our experiments with Keras library show that the precision error for the parameter recovery for the tested networks is less than $10^{-13}$ with the usage of 64-bit floats, which improves the current state of the art by six orders of magnitude.
Fault attacks are among the well-studied topics in the area of cryptography. These attacks constitute a powerful tool to recover the secret key used in the encryption process. Fault attacks work by forcing a device to work under non-ideal environmental conditions (such as high temperature) or external disturbances (such as glitch in the power supply) while performing a cryptographic operation. The recent trend shows that the amount of research in this direction—which ranges from attacking a particular primitive, proposing a fault countermeasure, to attacking countermeasures—has grown up substantially and is going to stay as an active research interest for the foreseeable future. Hence, it becomes apparent to have a comprehensive yet compact study of the (major) works. This work, which covers a wide spectrum in the present-day research on fault attacks that fall under the purview of the symmetric key cryptography, aims at fulfilling the absence of an up-to-date survey. We present mostly all aspects of the topic in a way that is not only understandable for a non-expert reader, but also helpful for an expert as a reference.
Laser fault injection is one of the strongest fault injection techniques. It offers a precise area positioning and a precise timing, allowing a high repeatability of experiments. In our paper we examine possibilities of laser-induced faults that could lead to instruction skips. After the profiling phase we were able to perform an attack on the last AddRoundKey operation in AES and to retrieve the secret key with just one faulty and correct ciphertext pair. Our experiments show very high degree of repeatability and 100% success rate with correct laser settings.
16 November 2015 in Wireless Personal Communications