Machine Learning Based IoT Edge Node Security Attack and Countermeasures

Abstract

Advances in technology have enabled tremendous progress in the development of a highly connected ecosystem of ubiquitous computing devices collectively called the Internet of Things (IoT). Ensuring the security of IoT devices is a high priority due to the sensitive nature of the collected data. Physically Unclonable Functions (PUFs) have emerged as critical hardware primitive for ensuring the security of IoT nodes. Malicious modeling of PUF architectures has proven to be difficult due to the inherently stochastic nature of PUF architectures. Extant approaches to malicious PUF modeling assume that a priori knowledge and physical access to the PUF architecture is available for malicious attack on the IoT node. However, many IoT networks make the underlying assumption that the PUF architecture is sufficiently tamper-proof, both physically and mathematically. In this work, we show that knowledge of the underlying PUF structure is not necessary to clone a PUF. We present a novel non-invasive, architecture independent, machine learning attack for strong PUF designs with a cloning accuracy of 93.5% and improvements of up to 48.31% over an alternative, two-stage brute force attack model. We also propose a machine-learning based countermeasure, discriminator, which can distinguish cloned PUF devices and authentic PUFs with an average accuracy of 96.01%. The proposed discriminator can be used for rapidly authenticating millions of IoT nodes remotely from the cloud server.

Socio-network Analysis of RTL Designs for Hardware Trojan Localization

“…Then, we discard node (9,10) from further consideration as it is reachable from the rest of the nodes (C=0.4). As higher CC indicates maximum closeness, we sort the CC column and find that the nodes (9,12) and (8,11) both have minimum CC as well as moderate closeness (C=0.2). However, for BC, we see the value of each node is '0'.…”

“…During explicit payload, we use existing nets (8,11) and (9,12) to act as HT triggering signal (HTT), and the edge between N12 and N13 is being modified to work as HT payload (HTP) due to the inclusion of output of HTT. In comparison to Fig.…”

“…For implicit payload, we add an edge between nodes (N8 and N9), which would act as input to HTT with other input being the edge between nodes (N9 and N12). However, we have four choices to create additional nodes between (8,11) and (9,12) who can act as HTT, as shown in Table III. Among four possibilities, we choose node (8,9) as an additional HTT signal.…”

“…We propose an algorithm as shown above to locate nodes that can act as combinational HT. The design is considered to (8,9) Impact both outputs' paths without any delay difference (8,12) Impact one output and present path anomaly (11,12) Create a back-edge (9,11) Reachable by visiting node N10 be represented as DAG and edge-labeled. We describe major procedures in the algorithm as follows:…”

Machine Learning Attacks and Countermeasures for PUF-Based IoT Edge Node Security

“…We also show that the approach can be extended, through unsupervised noisy pretraining to handle two (2) standard encryption protocols and three (3) common PUF architectures, which form some of the more common node authentication setups in practice. The preliminary version of this work has been published in [19,26]. The rest of this paper is organized as follows.…”