The Impact of Run-Time Variability on Side-Channel Attacks Targeting FPGAs

Davide Galli, Adriano Guarisco, William Fornaciari, Matteo Matteucci, Davide Zoni

Abstract

To defeat side-channel attacks, many recent countermeasures work by enforcing random run-time variability to the target computing platform in terms of clock jitters, frequency and voltage scaling, and phase shift, also combining the contributions from different actuators to maximize the side-channel resistance of the target. However, the robustness of such solutions seems strongly influenced by several hyper-parameters for which an in-depth analysis is still missing. This work proposes a fine-grained dynamic voltage and frequency scaling actuator to investigate the effectiveness of recent desynchronization countermeasures with the goal of highlighting the link between the enforced run-time variability and the vulnerability to side-channel attacks of cryptographic implementations targeting FPGAs. The analysis of the results collected from real hardware allowed for a comprehensive understanding of the protection offered by run-time variability countermeasures against side-channel attacks.

Introduction

Side-channel attacks (SCA) represent a significant threat to hardware security, allowing attackers to extract cryptographic keys by monitoring physical leakages such as power consumption or electromagnetic emissions. To mitigate these threats, designers employ "hiding" countermeasures that aim to reduce the signal-to-noise ratio or desynchronize the leakage traces.

Dynamic Voltage and Frequency Scaling (DVFS) has emerged as a popular hiding technique, introducing run-time variability to thwart Correlation Power Analysis (CPA) and Template Attacks (TA). However, the specific impact of different variability "knobs"—voltage, frequency, and clock phase—has remained under-explored. This study investigates these parameters using a custom-built actuator on FPGA hardware to determine which factors truly contribute to side-channel resistance.

Experimental Methodology

The researchers developed a specialized hardware module called rDVFS (random DVFS) for AMD Xilinx FPGAs. A primary challenge in implementing DVFS is that standard frequency changes often require stopping the clock, which can lead to system instability. The rDVFS module overcomes this using a "master-slave" configuration with two Mixed-Mode Clock Managers (MMCMs). While one MMCM drives the system, the other reconfigures in the background, allowing for seamless, continuous operation during transitions.

Testbed and Data Collection

The evaluation targeted an AES-128 implementation running on a RISC-V soft-core. The experiments were conducted on real hardware using:

  • Hardware: AMD Xilinx Artix-7 FPGAs (100T and 35T variants) on a NewAE CW305 board.
  • Data: Up to 100,000 power traces per scenario, captured at 250 MSamples/s.
  • Attack Vectors: Correlation Power Analysis (CPA) and Template Attacks (TA).

Key Findings

The Voltage Scaling "Mirage"

One of the most significant findings is that voltage scaling provides a deceptive sense of security. While randomizing voltage levels initially lowered the correlation coefficient ($\rho$), the researchers discovered that this was primarily due to low-frequency noise generated by the analog components of the voltage regulator. By applying a simple high-pass filter (HPF) to the collected traces, the researchers were able to strip away this noise, making the underlying secret leakage visible again. Consequently, voltage scaling alone is an insufficient defense against a sophisticated attacker.

Frequency Scaling: Range over Granularity

In contrast to voltage scaling, frequency randomization proved highly effective. The study challenged the common assumption that a higher number of frequency steps leads to better security. Instead, the results indicate that the frequency range (the spread between the minimum and maximum frequencies) is the critical factor for desynchronization. In scenarios with a wide frequency range (e.g., 25 MHz to 75 MHz), both CPA and Template Attacks failed to recover the key even after 100,000 traces.

Phase Shifting and Inter-chip Variability

The study also evaluated clock phase shifting and inter-chip manufacturing variability. Randomly shifting the clock phase provided negligible protection against the evaluated attacks. Furthermore, the researchers confirmed that inter-chip variability is not a reliable defense; a template model trained on one FPGA chip could successfully extract keys from a different chip of the same model.

Conclusion and Future Work

The research demonstrates that while run-time variability is a potent tool for FPGA security, not all actuators are created equal. Frequency scaling is a robust defense, provided the range is sufficiently wide, whereas voltage scaling and phase shifting offer limited protection.

The primary drawback of effective frequency scaling is the performance trade-off, as fluctuating frequencies result in unpredictable and generally slower execution times. Future research should focus on optimizing the "sweet spot" between security and performance, identifying the minimum frequency range required to maintain high resistance while minimizing system impact. Additionally, extending this analysis to modern UltraScale+ architectures remains a priority for future hardware security assessments.

Summary updated

Abstract

To defeat side-channel attacks, many recent countermeasures work by enforcing random run-time variability to the target computing platform in terms of clock jitters, frequency and voltage scaling, and phase shift, also combining the contributions from different actuators to maximize the side-channel resistance of the target. However, the robustness of such solutions seems strongly influenced by several hyper-parameters for which an in-depth analysis is still missing. This work proposes a fine-grained dynamic voltage and frequency scaling actuator to investigate the effectiveness of recent desynchronization countermeasures with the goal of highlighting the link between the enforced run-time variability and the vulnerability to side-channel attacks of cryptographic implementations targeting FPGAs. The analysis of the results collected from real hardware allowed for a comprehensive understanding of the protection offered by run-time variability countermeasures against side-channel attacks.
BibTeX
@misc{galli2024impact,
  author = {Davide Galli and Adriano Guarisco and William Fornaciari and Matteo Matteucci and Davide Zoni},
  title = {{The Impact of Run-Time Variability on Side-Channel Attacks Targeting FPGAs}},
  year = {2024},
  eprint = {2409.01881},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2409.01881},
}
Figure