Attack on a PUF-based Secure Binary Neural Network
By Bijeet Basak, Nupur Patil, Kurian Polachan, Srinivas Vivek
International Institute of Information Technology, Bangalore, India
Abstract
Binarized Neural Networks (BNNs) deployed on memristive crossbar arrays provide energy-efficient solutions for edge computing but are susceptible to physical attacks due to memristor nonvolatility. Recently, Rajendran et al. (IEEE Embedded Systems Letter 2025) proposed a Physical Unclonable Function (PUF)-based scheme to secure BNNs against theft attacks. Specifically, the weight and bias matrices of the BNN layers were secured by swapping columns based on device's PUF key bits.
In this paper, we demonstrate that this scheme to secure BNNs is vulnerable to PUF-key recovery attack. As a consequence of our attack, we recover the secret weight and bias matrices of the BNN. Our approach is motivated by differential cryptanalysis and reconstructs the PUF key bit-by-bit by observing the change in model accuracy, and eventually recovering the BNN model parameters. Evaluated on a BNN trained on the MNIST dataset, our attack could recover 85% of the PUF key, and recover the BNN model up to 93% classification accuracy compared to the original model's 96% accuracy. Our attack is very efficient and it takes a couple of minutes to recovery the PUF key and the model parameters.
To read the full article, click here
Related Semiconductor IP
- Digital PUF IP
- PUF IP
- PUF FPGA-Xilinx Premium with key wrap
- PUF FPGA-Xilinx Base
- ASIL-B Ready PUF Hardware Premium with key wrap and certification support
Related White Papers
- Input-Triggered Hardware Trojan Attack on Spiking Neural Networks
- How a voltage glitch attack could cripple your SoC or MCU - and how to securely protect it
- Neural Networks Can Help Keep Connected Vehicles Secure
- Interstellar: Fully Partitioned and Efficient Security Monitoring Hardware Near a Processor Core for Protecting Systems against Attacks on Privileged Software
Latest White Papers
- Attack on a PUF-based Secure Binary Neural Network
- BBOPlace-Bench: Benchmarking Black-Box Optimization for Chip Placement
- FD-SOI: A Cyber-Resilient Substrate Against Laser Fault Injection—The Future Platform for Secure Automotive Electronics
- In-DRAM True Random Number Generation Using Simultaneous Multiple-Row Activation: An Experimental Study of Real DRAM Chips
- SPAD: Specialized Prefill and Decode Hardware for Disaggregated LLM Inference