Physical Unclonable Function-based Key Sharing for IoT Security

Jiliang Zhang, Gang Qu

Abstract

In many Industrial Internet of Things (IIoT) applications, resource constraints prevent the use of traditional cryptographic solutions. Silicon Physical Unclonable Functions (PUFs) offer a lightweight alternative by leveraging manufacturing variations for key generation. However, standard PUFs are device-unique, making key sharing difficult. This paper proposes a novel Crossover Ring Oscillator (CRO) PUF architecture that utilizes interstage crossing structures and machine learning modeling to enable different devices to generate identical shared keys, facilitating secure communication in resource-constrained environments.

Introduction

The proliferation of the Industrial Internet of Things (IIoT) has introduced a vast array of low-power devices that lack the computational resources for standard cryptographic primitives like RSA or ECC. Physical Unclonable Functions (PUFs) have emerged as a promising solution, providing hardware-based security by exploiting unique physical characteristics of chips. While excellent for device identification, the inherent uniqueness of PUFs has historically limited their use in protocols requiring shared secrets. This research introduces a method to overcome this limitation using a flexible hardware architecture and a trusted coordination protocol.

The Crossover Ring Oscillator (CRO) PUF

The core innovation is the Crossover Ring Oscillator (CRO) PUF. Unlike traditional RO PUFs where paths are fixed, the CRO PUF incorporates interstage crossing structures implemented via Lookup Tables (LUTs). These structures allow signals to be routed between different inverter levels based on configuration bits, creating an exponential number of possible delay paths.

Architecture of the proposed CRO PUF showing interstage crossing levels and challenge-response mechanism.

By adjusting the configuration bits ($S_1, S_2, ...$), the physical path of the oscillator can be customized. This flexibility provides the mathematical "room" needed to find overlapping output characteristics across different physical chips.

Key-Sharing Protocol and ML Modeling

To enable two distinct devices (Alice and Bob) to generate the same key, a Trusted Third Party (TTP) is employed. The process involves several stages:

  1. Characterization: The TTP uses machine learning to build a precise delay model (matrix) for each device's PUF.
  2. Challenge Generation: Given a desired shared key $K$, the TTP uses its models to find specific challenges ($C_A$ for Alice and $C_B$ for Bob) that will result in the same output bitstream on their respective hardware.
  3. Key Generation: Alice and Bob receive their unique challenges and run them through their CRO PUFs to locally generate the identical key $K$.
The proposed key-sharing protocol involving Alice, Bob, and a Trusted Third Party (TTP).

Experimental Results

The proposed system was validated using both large-scale simulations and hardware implementations on Xilinx ZedBoard FPGAs.

  • Accuracy: Machine learning models achieved near 100% accuracy in predicting delay matrices, ensuring that the TTP can reliably find matching challenges.
  • Efficiency: The CRO PUF architecture is significantly more efficient than previous attempts at shareable PUFs. A 4×5 CRO PUF requires only 62 LUTs, representing an 11.54x reduction in hardware overhead compared to traditional neighbor-coding methods.
  • Speed: Training the ML model for a 4×5 CRO PUF takes less than half a second on standard hardware, making the system scalable for large device deployments.

Security and Reliability

The research addresses environmental stability through a "reliability threshold" algorithm, ensuring that the generated keys remain consistent despite temperature fluctuations or voltage noise. Formal verification using the Scyther tool was conducted to confirm protocol security against standard interception and impersonation attacks. However, the security model assumes a Trusted Third Party; if the TTP's database of device delay matrices is compromised, the security of the connected devices is at risk.

Conclusion

The CRO PUF-based key sharing method provides a viable path for implementing secure communication in the most resource-constrained IIoT environments. By shifting the computational burden of key coordination to a trusted server and utilizing a highly flexible hardware primitive, the system achieves a balance of security, efficiency, and functionality previously unavailable to PUF-based systems.

Summary updated

Abstract

In many Industry Internet of Things (IIoT) applications, resources like CPU, memory, and battery power are limited and cannot afford the classic cryptographic security solutions. Silicon Physical Unclonable Function (PUF) is a lightweight security primitive that exploits manufacturing variations during the chip fabrication process for key generation and/or device authentication. However, traditional weak PUFs such as Ring Oscillator (RO) PUF generate chip-unique key for each device, which restricts their application in security protocols where the same key is required to be shared in resource-constrained devices. In order to address this issue, we propose a PUF-based key sharing method for the first time. The basic idea is to implement one-to-one input-output mapping with Lookup Table (LUT)-based interstage crossing structures in each level of inverters of RO PUF. Individual customization on configuration bits of interstage crossing structure and different RO selections with challenges bring high flexibility. Therefore, with the flexible configuration of interstage crossing structures and challenges, CRO PUF can generate the same shared key for resource-constrained devices, which enables a new application for lightweight key sharing protocols.
BibTeX
@misc{zhang2018physical,
  author = {Jiliang Zhang and Gang Qu},
  title = {{Physical Unclonable Function-based Key Sharing for IoT Security}},
  year = {2018},
  eprint = {1807.10884},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1807.10884},
}
Figure