Abstract:
In today’s interconnected world, a vast number of computer systems are globally connected, creating a global village. These systems are remotely accessible and share a
significant amount of data on the cloud, raising concerns about data and system security. Field Programmable Gate Array (FPGA) technology has gained popularity due to its
fast computation, reconfigurability, and post-manufacturing reprogrammability. FPGAs
are built on current semiconductor technologies, making them susceptible to disturbances
from alterations in their fabrication process and runtime conditions. These variations can
have security implications that are not extensively explored.
In our research project, we investigated potential security issues related to side-channel
attacks (SCAs) on FPGAs and explored possible countermeasures. Firstly, we focused on
power analysis or power profiling of FPGAs, which rely on measuring voltage fluctuations during encryption tasks. These voltage fluctuations in the cryptographic module can
be measured using physical sources like an oscilloscope or remote sources like delay line
sensors. Secondly, we delved into power analysis-based SCAs that leverage these voltage
measurements to extract the secret key. Thirdly, we devised a framework based on machine learning and deep learning algorithms to predict secret keys and execute successful
attacks. Our custom CNN model outperformed previous studies, achieving a significant
improvement of approximately 46% by successfully attacking with only 570 attack power
traces. Fourthly, we explored state-of-the-art resilient countermeasures against power
analysis-based SCAs on FPGAs and identified the hiding technique as the most effective
one.
Looking ahead, these attacks are not limited to individual FPGAs. Cloud FPGAs and IoT
devices are also susceptible to power analysis attacks, exploiting partial or full access to
the power distribution networks (PDN). Therefore, addressing these security concerns is
crucial for ensuring the safety and integrity of FPGA-based systems and IoT devices in
the future.
