Resilient Endurance-Aware NVM-based PUF against Learning-based Attacks

TL;DR

The paper addresses the endurance limitations of NVM-based PUFs under learning-based attacks by introducing an analytical Markov-chain model to predict wear and lifetime. It then proposes REAP-NVM, an endurance-aware NVM-PUF that minimizes per-challenge writes (writing only one cell-pair) and exploits multilevel NVM cells to boost ML-resilience. Empirical results show REAP-NVM achieving about a 62× improvement in endurance over state-of-the-art designs while preserving strong ML-resilience (attack-prediction accuracy ~55%) and maintaining good quality metrics (Uniformity/Uniqueness near mid-range). The work demonstrates a practical pathway to secure, long-lived NVM-PUFs suitable for resource-constrained devices, with clear trade-offs in area and energy that can guide future optimizations.

Abstract

Physical Unclonable Functions (PUFs) based on Non-Volatile Memory (NVM) technology have emerged as a promising solution for secure authentication and cryptographic applications. By leveraging the multi-level cell (MLC) characteristic of NVMs, these PUFs can generate a wide range of unique responses, enhancing their resilience to machine learning (ML) modeling attacks. However, a significant issue with NVM-based PUFs is their endurance problem; frequent write operations lead to wear and degradation over time, reducing the reliability and lifespan of the PUF. This paper addresses these issues by offering a comprehensive model to predict and analyze the effects of endurance changes on NVM PUFs. This model provides insights into how wear impacts the PUF's quality and helps in designing more robust PUFs. Building on this model, we present a novel design for NVM PUFs that significantly improves endurance. Our design approach incorporates advanced techniques to distribute write operations more evenly and reduce stress on individual cells. The result is an NVM PUF that demonstrates a $62\times$ improvement in endurance compared to current state-of-the-art solutions while maintaining protection against learning-based attacks.

Figures (10)

• Figure 1: Flow to analyze the endurance of the PUF. Based on a Markov chain reconstruction of the PUF the distribution of set and reset operations can be deduced to infer when would the PUF be unusable.
• Figure 2: Markov chain modeling the behavior of cells in PUF from Arbiter_mpuf.
• Figure 3: Design of REAP-NVM. Similar to an AUF but with the addition of NVM cells with variable delay to increase the security.
• Figure 4: Markov chain modeling the behavior of cells in REAP-NVM. It enhances over the state-of-the-art design from Arbiter_mpuf by reducing the probability of set and reset per cell.
• Figure 5: Distribution of REAP-NVM's Uniformity and Uniqueness. Both have the desired Gaussian distribution centered around 64.