Autonomous vehicles (AVs) have promised to drastically improve the convenience of driving by releasing the burden of drivers and reducing traffic accidents with more precise control. With the fast development of artificial intelligence and significant advancements of the Internet of Things technologies, we have witnessed the steady progress of autonomous driving over the recent years. As promising as it is, the march of autonomous driving technologies also faces new challenges, among which security is the top concern. In this article, we give a systematic study on the security threats surrounding autonomous driving, from the angles of perception, navigation, and control. In addition to the in-depth overview of these threats, we also summarize the corresponding defense strategies. Furthermore, we discuss future research directions about the new security threats, especially those related to deep-learning-based self-driving vehicles. By providing the security guidelines at this early stage, we aim to promote new techniques and designs related to AVs from both academia and industry and boost the development of secure autonomous driving.

The Security of Autonomous Driving: Threats, Defenses, and Future Directions

/pdf/modern-engineering-statistics-5f8v8a41dp.pdf

Modern engineering statistics

With the ongoing development and improvement of deep learning technology, autonomous vehicles (AVs) have made tremendous progress in recent years. Despite its great potential, AV supported by deep learning technology still faces numerous security threats, which prevent AV from being putting into large-scale practice. Aiming at this challenging situation, in this paper, we would like to exploit two attacks against deep learning algorithms in traffic sign recognition system by leveraging particle swarm optimization. Specifically, we first exploit the PAPSO (poisoning attack with particle swarm optimization) which focuses on the training process of the deep learning algorithms in the traffic sign recognition system, i.e., the attacker injects crafted samples into the training dataset, causing a reduction in classification accuracy of the traffic sign recognition system. Then, we also explore the EAPSO (evasion attack with particle swarm optimization) which on the other hand focuses on the interference process of the deep learning algorithms, i.e., the attacker adds some hardly perceptible perturbations to the targeted test sample, leading to a misclassification on it. Extensive experiments are conducted to shed light on the effectiveness of our attacks, and some corresponding defense strategies are also presented.

Poisoning and Evasion Attacks Against Deep Learning Algorithms in Autonomous Vehicles

Performance of electromagnetic interference (EMI) filters in power-electronics applications is limited by the parasitic coupling between the components of the filters and the self-parasitic of each component. While these parasitic effects can be partially taken into account on the circuit level, it is difficult to estimate their values. In this article, a full-wave modeling methodology is proposed to predict the performance of a complete EMI filter up to 1 GHz. Following the proposed methodology, the mutual couplings among the EMI filter components are taken into account as well as the self-parasitics of each individual component. Experiments and simulations are carried out to validate the modeling methodology. A self-parasitic cancellation technique is also applied to demonstrate the benefits of three-dimensional modeling methodology in EMI filter design.

Modeling Strategy for EMI Filters

In this paper, test study on low-cost civil engineering construction material is presented. In fact, in several civil constructions the strategy is to build up to flrst \fence" to EMI that limits interferences while using only in some limited areas high-cost ad hoc shielding material. The materials used in this study are tested at the Universitµa di Napoli Parthenope, formerly Istituto Universitario Navale (IUN) Reverberating Chamber (RC) according to a nested approach. Tests are made in the frequency range from 2 to 18GHz and show that low-cost materials are able to achieve about 16dB of shielding efiectiveness (SE) in contrast to ad hoc materials that show about 50dB in SE.

/pdf/shielding-effectiveness-tests-of-low-cost-civil-engineering-3csriwag2w.pdf

Shielding effectiveness tests of low-cost civil engineering materials in a reverber- ating chamber

A new technique for shielding effectiveness measurements is the dual VIRC method. In this method two Vibrating Intrinsic Reverberation Chambers (VIRC) are combined together via a common wall with an aperture that forms the interface between them. This particular set-up makes it possible to achieve a very high dynamic range. Advantage of this technique compared to the nested reverberation set-up is that measurements can be conducted much faster and at much lower frequency. Several different approaches to determine the shielding effectiveness are presented. It has been shown that multiple interactions play a role in the particular set-up.

Vibrating Intrinsic Reverberation Chambers for shielding effectiveness measurements

Shielding effectiveness measurements in a dual vibrating intrinsic reverberation chamber results in very repeatable results for a large frequency range and large dynamic range. Antennas in reverberation chambers do not need any gain, but the losses should be low and the dimensions should be small. The conical and discone antenna are very suitable for this purpose. An optimized conical and a discone antenna have been designed, modeled and simulated and built. Experiments in free space show their wideband behavior. Measurements inside the reverberation chamber have been performed as well.

Wide-band antennas for reverberation chambers

Cars equipped with remote keyless entry systems are very vulnerable to electromagnetic interference because of the wireless link. Carrying the encrypted signal between the key fob and the car, it can deliberately be jammed by a malicious person. Jamming on the physical layer leaves the encryption and code algorithms on the higher communication layers useless. Jamming for criminal purposes is considered to be intentional electromagnetic interference. In this paper, the robustness of the remote keyless entry system to intentional electromagnetic interference is investigated. Two commercially available remote keyless entry systems are analyzed. Measurements are performed in an anechoic chamber to estimate the robustness of the RF receivers. Both systems are equipped with super regenerative receivers. Results show that the selectivity of both systems are poor and they are easily jammed. Field strenghts as low as 0.1 V/m can already jam the wireless link over a wide frequency bandwidth of 4 MHz. A super heterodyne receiver would be robuster to electromagnetic interference than a super regenerative receiver, but this type of receiver has a relative high cost.

Robustness of remote keyless entry systems to intentional electromagnetic interference

A modern hospital has a wide range of medical devices that can be technologically very advanced and complex in terms of electromagnetic emissions and susceptibility. It is a public environment where people carry communication equipment and/or medical active implants. Therefore, electromagnetic interference among the devices in the hospital environment is not an unfamiliar topic. There is a misconception among engineers that if all medical devices comply with harmonized standards, then the installation of a large number of medical devices altogether would work absolutely fine. However, this is not always true, and so for complex systems, a smarter approach is necessary to assess and control the risks of electromagnetic interference. This paper emphasizes the importance of using the risk-based approach to deal with the often unpredictable factors causing electromagnetic interference in a complex hospital environment.

Risk-based EMC Approach in Hospital Environment

The theoretical field strength values in reverberant environments can easily be approximated using the well-established models. In the near-average range of the distribution a very good agreement can be reached and is used in many applications. The estimation of the maximal field strengths is, however, difficult due to certain assumptions. The complexity of real reverberant cavities is too high for analytical and even numerical approach. This study focuses on the experimental way using a Vibrating Intrinsic Reverberation Chamber (VIRC). The randomness of movement of flexible walls of a VIRC is exploited by conducting long, 1 hour measurements to obtain many independent samples. Statistical analysis of the data is conducted to evaluate the maximal, average, and their ratio as functions of both number of samples and measurement time.

Robert Vogt-Ardatjew

Papers

Vibrating Intrinsic Reverberation Chambers for shielding effectiveness measurements

Wide-band antennas for reverberation chambers

Robustness of remote keyless entry systems to intentional electromagnetic interference

Risk-based EMC Approach in Hospital Environment

Observation of maximal and average field values in a Reverberation chamber