As LoRaWAN is one of the most popular long-range wireless protocols among low-power IoT applications, more and more focus is shifting towards security. In particular, physical layer topics become relevant to improve the security of LoRaWAN nodes, which are often limited in terms of computational power and communication resources. To this end, e.g., detection methods for wireless attacks improve the integrity and robustness of LoRaWAN access. Further, wireless physical layer techniques have potential to enhance key refreshment and device authentication. In this work, we aim to provide a comprehensive review of various vulnerabilities, countermeasures and security enhancing features concerning the LoRaWAN physical layer. Afterwards, we discuss the impact of the reviewed topics on LoRaWAN security and, subsequently, we identify research gaps as well as promising future research directions.
The low-power wide-area (LPWA) technologies, which enable cost and energy-efficient wireless connectivity for massive deployments of autonomous machines, have enabled and boosted the development of many new Internet of things (IoT) applications; however, the security of LPWA technologies in general, and specifically those operating in the license-free frequency bands, have received somewhat limited attention so far. This paper focuses specifically on the security and privacy aspects of one of the most popular license-free-band LPWA technologies, which is named LoRaWAN. The paper’s key contributions are the details of the design and experimental validation of a security-focused testbed, based on the combination of software-defined radio (SDR) and GNU Radio software with a standalone LoRaWAN transceiver. By implementing the two practical man-in-the-middle attacks (i.e., the replay and bit-flipping attacks through intercepting the over-the-air activation procedure by an external to the network attacker device), we demonstrate that the developed testbed enables practical experiments for on-air security in real-life conditions. This makes the designed testbed perspective for validating the novel security solutions and approaches and draws attention to some of the relevant security challenges extant in LoRaWAN.
We present a watermarking scheme for sensor data, which accurately detects sensor voltage manipulation attacks. Due to a low complexity implementation running locally on a microcontroller, the proposed technique is well suited for home automation systems but also IoT systems where lightweight sensor nodes with restricted communication capabilities are commonly applied. The watermarking technique is based on modulating the sensor supply voltage, which embeds the watermark signal into the sensor output voltage. Hence, invasive voltage tampering attempts including ADC grounding and voltage injection attacks can be efficiently detected. According to the performed characterization results, our method is able to distinguish normal operation and an attack condition with approx. 98% accuracy for the entire ADC input voltage range. In a real-world demonstration involving a thermistor based temperature sensor all attack attempts were successfully detected.
This paper presents a comprehensive investigation of machine learning-based intrusion detection methods to reveal cyber attacks in railway axle counting networks. In contrast to the state-of-the-art works, our experimental results are validated with testbed-based real-world axle counting components. Furthermore, we aimed to detect targeted attacks on axle counting systems, which have higher impacts than conventional network attacks. We present a comprehensive investigation of machine learning-based intrusion detection methods to reveal cyber attacks in railway axle counting networks. According to our findings, the proposed machine learning-based models were able to categorize six different network states (normal and under attack). The overall accuracy of the initial models was ca. 70–100% for the test data set in laboratory conditions. In operational conditions, the accuracy decreased to under 50%. To increase the accuracy, we introduce a novel input data-preprocessing method with the denoted gamma parameter. This increased the accuracy of the deep neural network model to 69.52% for six labels, 85.11% for five labels, and 92.02% for two labels. The gamma parameter also removed the dependence on the time series, enabled relevant classification of data in the real network, and increased the accuracy of the model in real operations. This parameter is influenced by simulated attacks and, thus, allows the classification of traffic into specified classes.
This paper focuses on the evaluation of pulse drivers for the use in class-S power amplifier setups with respect to linearity. Nonlinear effects are likely to degrade the overall system performance, though they are hard to characterize especially in an early stage of design. Based on measurements conducted with a configurable pulse driver, namely Triquint's TGA4954-SL, a simple gauging method was developed. It allows the identification of favorable bias-parameter settings or the comparison of different devices while requiring only very little knowledge of the overall system.