In this article, we introduce a novel approach to achieving lightweight device authentication through the use of a low-complexity Convolutional Neural Network (CNN). In our work, we improve the False Authentication Rate (FAR) by transforming the standard CNN into a Bayesian CNN (BCNN or BNN). This transformation enables the use of probabilistic modelling techniques, increasing the model’s robustness and its confidence in authentication decisions. Regardless of the model used, clients authenticate with a retention-based Dynamic Random Access Memory Physical Unclonable Function (DRAM PUF) response. Our approach integrates the low computational complexity of the CNN with the intrinsic security characteristics of the DRAM PUF, offering a robust solution for lightweight and secure device authentication.

In this work, we analyse Commercial Off-The-Shelf (COTS) Resistive Random Access Memory (ReRAM) modules for their suitability to implement a novel runtime-accessible True Random Number Generator (TRNG). For this purpose, modules from two different manufacturers (Adesto Technologies and Fujitsu) were tested, which exhibited distinct characteristics under different conditions. If suitable parameters are selected, the proposed TRNG can successfully pass all the tests of both the NIST SP800-22 Statistical Test Suite and the NIST SP800-90B Entropy Source Test Suite at a wide range of temperatures. At the same time, the TRNG achieves a throughput of at least 28 bits per second under adverse temperature conditions and approximately 51 bits per second at room temperature, in the worst case. Therefore, the performance of the TRNG is sufficient for many practical applications such as security protocols for the Internet of Things (IoT) and in-vehicle networks [1], [2].

In this work, we present a novel method of increasing the entropy of the CNT-PUF, a Physical Unclonable Function (PUF) based on Carbon-NanoTube Field Effect Transistors (CNT-FETs). The binary responses of this PUF are based on the drain current IDof each CNT-FET under the influence of a particular gate-source voltage VGS,which, through the employment of a single threshold value for ID,can indicate whether each relevant CNT cell of the array is conducting (acting either as a true conductor or as a semiconductor) or not (acting as an insulator). In this work, we propose the adoption of individual threshold values for each such cell as part of the relevant PUF challenge, thereby significantly increasing the overall entropy of this PUF, as well as the security that it can provide. Moreover, this method allows for the realisation of a source of higher entropy in the form of a True Random Number Generator (TRNG). Finally, we note that our work and its results are most probably also relevant for other CNT- based PUFs, structures, and primitives that utilise a single current (or even, voltage) threshold to determine the state of the different CNT cells utilised.

Physical Unclonable Functions (PUFs) are increasingly used in the process of securing applications. For this purpose, it is crucial that the PUF satisfies all the required properties adequately, including Unpredictability. An important aspect of Unpredictability is Randomness, which includes being free of spatial correlation effects. However, most methods for assessing randomness are not capable of detecting correlation, such that this aspect is often ignored. This work summarises the current literature to shed more light on the topic of analysing spatial correlation in weak PUFs, and evaluates the various methods proposed in the literature for detecting such effects. Additionally, the spatial correlation of a Dynamic Random Access Memory (DRAM) decay-based PUF implemented on the DRAM of a Raspberry Pi board, as well as that of a Carbon-NanoTube-based PUF (CNT-PUF), are examined, using, for the first time in the context of PUFs, not only other well-known metrics proposed in the relevant literature, but also the Getis-Ord G metric. Finally, a mitigation technique against attacks based on spatial auto-correlation is proposed and its effective application to PUF responses is discussed.

In recent years, a new generation of the Internet of Things (IoT 2.0) is emerging, based on artificial intelligence, the blockchain technology, machine learning, and the constant consolidation of pre-existing systems and subsystems into larger systems. In this work, we construct and examine a proof-of-concept prototype of such a system of systems, which consists of heterogeneous commercial off-the-shelf components, and utilises diverse communication protocols. We recognise the inherent need for lightweight security in this context, and address it by employing a low-cost state-of-the-art security solution. Our solution is based on a novel hardware and software co-engineering paradigm, utilising well-known software-based cryptographic algorithms, in order to maximise the security potential of the hardware security primitive (a Physical Unclonable Function) that is used as a security anchor. The performance of the proposed security solution is evaluated, proving its suitability even for real-time applications. Additionally, the Dolev-Yao attacker model is considered in order to assess the resilience of our solution towards attacks against the confidentiality, integrity, and availability of the examined system of systems. In this way, it is confirmed that the proposed solution is able to address the emerging security challenges of the oncoming era of systems of systems.