This PDF file contains the front matter associated with SPIE Proceedings Volume 11009 including the Title Page, Copyright information, Table of Contents, Introduction, and Conference Committee listing.

Unmanned aerial vehicles (UAVs) on patrol that recognize scene changes are a powerful instrument in a lot of surveillance and reconnaissance applications, e.g., protection of convoys from improvised explosive devices (IEDs) by early detecting deployment traces, intruder or manipulation detection for real estates and military camps, or danger recognition and assessment in disaster monitoring. For that purpose, a visual-optical camera records video streams in which an automatic video change detection computes relevant or suspicious changes between two patrols. The main challenge is to identify true scene changes, i.e. to filter out changes caused by camera brightness variations, illumination changes due to cloudiness or time of day, or varying shadows cast by clouds or scene objects. Different partly overlapping or orthogonal solutions for all that, like chromatic adaptation, local mean adjustment, retinex, intensity equalization, and CIE Lab shadow removal, are evaluated and advantages/disadvantages are compared. Finally, the optimal method selection and combination is determined in order to maximize the change detection performance. Depicted results document the reached progress.

The advantages of using far-infrared technology in guiding systems, warning systems or surveillance systems have been demonstrated in defense applications through a large span of imaging functions such as detection, target recognition and positive identification in complex environments. Used as a complementary tool in a set of ADAS sensing technologies, far-Infrared thermal camera provides the key sense to improve system performance and increase detection robustness. This enhanced capability relies on the detection of the unique thermal signature of a pedestrian, an animal or any obstacle in any weather and light conditions, day or night, without being glared by the sun or any light source. Implementing ADAS based on far-infrared solution into an autonomous vehicle level 3 and above can reduce the rate of false positive detection. Being an imaging technology, farinfrared camera using ULIS technology can be easily integrated into standard ADAS platform while minimizing computation resources required by detection algorithms thanks to the thermal signature detection.

Wearable technology is growing exponentially and becoming as part of our life where daily activities will be tracked and monitored. As more and more wearable device are connected to the internet, the more urgent the need for authentication is required. Biometrics is rapidly gaining popularity as a powerful authenticator to meet this challenge. Biometric technology enables users to identify themselves quickly and securely. However, because of the nature of the IoT device, there is a need for continuous authentication. Cardiovascular biometric technology such as ECG and PPG are already moving forward as a biometric continuous authentication. However, it was recently shown that an ECG signal is vulnerable to presentation attack. Since PPG is widely used in wearable devices, they are more vulnerable to presentation attack. In this paper, we introduce a systematic presentation attack on PPG biometric where a short template of the victim’s PPG is collected by an attacker and used to map the adversarial’s PPG into the victims.

Micro aerial vehicles (MAVs), in particular, quadcopters, have seen increasing adoption in many application domains due to their flexible operations and versatile functions. By taking advantages of the inherent system inertia properties of the physical platform during MAV operations, satisfactory flight performance of the vehicle can be achieved despite a variety of internal and external disturbances including environment perturbations and cyber-physical attacks, provided the fault detection and control strategy is maintained with a proper lower bound of its operating frequency. From the control design perspective, such frequency bound depends heavily on the sensitivity of the system response to its dynamic properties including inertia and dynamic couplings. For a cyber-physical system, which is subject to various cyber and physical attacks, there is usually a large sensitivity difference between physical time scale and cyber time scale. With this fact, a properly designed fault-tolerant and control strategy can leverage the inertia property of the system to arrive at an optimal control frequency to save computational power while maintaining the flight performance. For many current MAVs, their control programs are usually executed at a fixed high-frequency, which provides no significant performance gains at most of the time. In this work, we perform a sensitivity analysis and quantification of the vehicle system to guide the design and optimization of customized on-board control and computational resources. By quantifying system sensitivity and its variation across different control channels (DOFs), the computational resources can be re-distributed and optimized to detect and reject cyber-physical attacks more efficiently. With the proposed method, the cyber and physical resources of the vehicle can be allocated by system demands, which can further enhance vehicle reliability and accomplish complicated security tasks. To demonstrate the effectiveness of the proposed method, we have conducted the simulated flight test with the real vehicle parameters.

Physical Unclonable Functions (PUFs) act as functions encoded in hardware, which produce a unique output, being referred to as a response, for a specific input, being called a challenge. PUFs provide a varying level of security, and can, therefore, be used in different applications, depending on the number of their available inputoutput pairs, which are referred to as Challenge-Response Pairs (CRPs). For example, a PUF with only a single challenge-response pair can be used for identification, while a PUF with multiple CRPs can be used to provide multiple different session keys for authentication.^{1, 2} In the first case, the response needs to be secret, while, in the second one, responses can be also used without any secrecy, as long as the related CRPs are not used again. Generally, PUFs are vulnerable to modeling and machine learning attacks. In this paper we investigate and show the resiliency of DRAM-based PUFs against Machine Learning (Naive Bayes (NB), Logistic Regression (LR) and Support Vector machine (SVM)) and also Deep Learning (in particular convolutional neural network (CNN) attacks. We are the first to provide a detailed analysis of on-board DRAM startup values for the purpose of generating unique IDs and their vulnerabilities to attacks. We performed our experiments on the Digilent Atlys board (Xilinx Spartan 6 FPGA); using the on-board DRAM memories (MIRA P3R1GE3EGF G8E DDR2). Our results indicates that the 3 startup value-based DRAM PUFs (DRAM1, DRAM2, and DRAM3) are robust against machine learning attacks.

In a human operated vehicle, the alignment of tires aims to strike a balance between ease of steering and a minimization of tire wear. The replacement of the human driver in an autonomous vehicle with low latency computer control of path tracking means that tire alignment can be performed with less emphasis on handling characteristics which contribute to ease of steering and directed towards improvement in tire life. This study uses MATLABs Vehicle Dynamics Blockset and Predictive Driver block to compare the path tracking capability of a passenger vehicle performing a double lane change maneuver under the control of the pure pursuit autonomous path following algorithm as well as a simulated human driver. Validation of the Predictive Driver block is performed by tracking a panel of human drivers performing the double lane change maneuver using GPS for localization in a subcompact electric vehicle. The vehicle model is characterized based on measurements from the test vehicle and sent through the same double lane change in simulation to compare behaviors. Tire alignment parameters are altered to demonstrate their effects on vehicle handling under both types of vehicle control. In the simulation environment, the pure pursuit algorithm tracks the desired path consistently across all parameter variations while the simulated human driver varies in its path tracking capabilities.

There is a strong disconnect between cybersecurity and small to medium sized organizations that are members of Industry 4.0. Industry 4.0 (I4) is considered the 4th industrial revolution where automation and data exchange co-exist in manufacturing technologies. This next phase of industry includes cyber-physical systems, the Internet of things (IoT), cloud computing and cognitive computing. This report will survey three specific areas in relation to Industry 4.0 and its impact on the security of the supply chain: Automotive Cybersecurity, Industrial Internet of Things (IIoT) and Data Protection in the Manufacturing Industry.

Naval forces rely heavily on Cyber-Physical Systems (CPS) to conduct operations and achieve mission success. As CPS infrastructure becomes exposed to the contested world through networks, CPS security becomes much more important. In a CPS, the cyber components manage the physical components. ONR proposes that the overall goal for CPS resiliency is to have the physical systems behave properly regardless of fault and disruption. Our approach to CPS resiliency focuses on the physical components. We observed that the inertia of the physical components provides a natural but limited resilience, and is capable of tolerating short-term disruption without affecting the health and safety of the CPS. This and the fact that a CPS has a large difference between physical and cyber time scales, enables a unique approach to CPS resiliency. We present Byzantine Fault Tolerant++ (BFT++), a cyber resilient architecture that engineers the cyber components to be brittle against attack, which consequently forces cyber attacks and related disruptions to be short-lived and within tolerance of the physical system’s inertia.

The significant expansion of the Internet of Things (IoT) field and unique requirements of the IoT devices bring new technologies created exclusively to provide wireless connectivity for the IoT devices. Among these technologies, we can include LoRa technology. Unlike some other LPWANs, LoRa technology is an open standard, and it allows us to build private networks. We took advantage of that and developed our gateway. The paper deals with a proposal of network infrastructure and the hardware solution of the LoRaWAN gateway based on the second generation of monolithic microcomputer Raspberry Pi model B and fully compatible LoRaWAN 868 MHz iC880A concentrator. The concentrator is connected to Raspberry Pi via Serial Peripheral Interface (SPI). In 2018, five gateways were deployed to cover a nearly entire area of Ostrava city in the Czech Republic and its surrounding areas with LoRaWAN signal. Our solution uses The Things Network platform to connect to a global open crowd-sourced IoT data network. For the end-users comfort, we implemented a web application that serves as a backend for registration of the end-devices to the LoRaWAN network, and it provides access to the history of all uplink messages transmitted by end-devices and received by LoRaWAN network. The next part of the article discusses end-device proposed for network availability testing.

For autonomous vehicles 3D, rotating LiDAR sensors are often critically important towards the vehicle’s ability to sense its environment. Generally, these sensors scan their environment, using multiple laser beams to gather information about the range and the intensity of the reflection from an object. LiDAR capabilities have evolved such that some autonomous systems employ multiple rotating LiDARs to gather greater amounts of data regarding the vehicle’s surroundings. For these multi–LiDAR systems, the placement of the sensors determine the density of the combined point cloud. We perform preliminary research regarding the optimal LiDAR placement strategy on an off–road, autonomous vehicle known as the Halo project. We use the Mississippi State University Autonomous Vehicle Simulator (MAVS) to generate large amounts of labeled LiDAR data that can be used to train and evaluate a neural network used to process LiDAR data in the vehicle. The trained networks are evaluated and their performance metrics are then used to generalize the performance of the sensor pose. Data generation, training, and evaluation, was performed iteratively to perform a parametric analysis of the effectiveness of various LiDAR poses in the Multi–LiDAR system. We also, describe and evaluate intrinsic and extrinsic calibration methods that are applied in the multi–LiDAR system. In conclusion we found that our simulations are an effective way to evaluate the efficacy of various LiDAR placements based on the performance of the neural network used to process that data and the density of the point cloud in areas of interest.

In this paper, the attention is focused on the design of a decentralized Intelligent Transportation System (ITS) architecture in an Internet of Vehicle (IoV) environment. Introducing several independent Fog nodes, it is possible to take advantages of their computational properties for decreasing latency of the Centralized ITS Systems. Thus, a micro area controlled system is achieved. In this way, it will be possible to quickly and easily redesign ITS strategies. In this work, data coming from IoV layers are locally analyzed for checking anomalies by adopting mining algorithm. If something is going wrong, fog nodes locally face off the anomaly by taking a decision on incoming vehicle flow rates by choosing to limit or to completely block incoming traffic. In this work, we propose a cooperative strategy among decentralized ITS nodes for increasing safety and comfort of the drivers along their journeys. When local ITS policies fail, we propose a cooperative mechanism to find a macro-area optimized strategy. ITS node which detects issue will assume the role of master. Instead, neighbor fog nodes are involved by assuming the role of slave. Master node will decide policies and the slave node will apply these policies in the pertinent area. When issue is solved, the partnership is declared finished by the master. Each node returns master of their area. Moreover, this architecture is also based on a customized protocol for the IoV layer where Vehicle-to-Everythings (V2X) communications are optimized for assuring a good throughput and reducing interference.

For autonomous systems and vehicles to be able to traverse and path plan in their environment, location needs to be known. To determine location, two items are required, a map and a localization method. The purpose of this paper is to determine the differences and benefits between two low cost off the shelf Light Detection And Ranging (LiDAR) units in an unstructured environment for mapping and localization. The LiDAR units angularly sample the environment in a full 360-degree horizontal view, each with a varying sample density. The first has more vertical samples and a broader vertical field of view, where the other has fewer vertical samples and smaller vertical field of view but higher angular resolution. Also, both LiDAR units produce similar orders of magnitude of points per second. The two LiDAR units are compared with a simulation model of the test vehicle and LiDAR units as well as real world testing. The results of the experiments show that that mapping with a higher vertical resolution LiDAR unit appears to produce a better map, whereas localization with the higher angular resolution LiDAR unit produces more consistent results overall but runs considerably slower.

The positioning accuracy of global navigation satellite systems (GNSS) in dense urban canyon environment significantly deteriorates due to multipath and signal blockage. For this reason, inertial navigation system (INS) is often integrated with GNSS to ensure a reliable navigation solution during such periods of GNSS signal outages. A low-cost navigation solution for land vehicles has been developed by integrating GNSS positioning solution with the measurements from the vehicle motion sensors (accelerometers and gyroscopes). The major drawback of the usage of these inertial sensors is its progressive error accumulation, where the gyroscope drift errors increase gradually, leading to an unusable position estimate, especially in the absence of GNSS updates. Navigation in GNSS-denied environment requires aiding INS with other exteroceptive sensors such as cameras to guarantee the continuity of reliable positioning updates. The estimation of the camera’s relative change in position and orientation over time is known as visual odometry (VO). A VO-based multisensor integrated navigation system is presented here to surmount the inaccuracy of GNSS in urban scenarios and the drifts of the motion sensors. To enhance the overall system accuracy of the VO-based integrated solution, this paper explores improving the positioning accuracy during GNSS outages by nonlinear modeling of the residual position errors using a neural network. The results show a significant accuracy improvement over relatively long GNSS outages.

A method of integrating non-automated vehicles into a platoon of Cooperative Adaptive Cruise Control (CACC) Class 8 Heavy Duty Trucks was developed and tested at the National Center for Asphalt Technology (NCAT) test track. As CACC platooning becomes more commonplace, the interaction between automated and non-automated vehicles increases. This work explores this interaction during a highway merging scenario. Merging of vehicles occurs when a vehicle enters the highway system where an automated platoon is operating at that time. This scenario is handled by predicting the trajectories of the automated platoon and the non-automated vehicle entering the highway. Based on these trajectories, the automated platoon decides to create a gap for the merging vehicle if necessary. In this work, the non-automated vehicle is assumed to be a passive, connected vehicle, i.e. the platoon has knowledge of the merging vehicle’s position and velocity. The merging scenarios are handled by a CACC platooning system with Dedicated Short Range Communication (DSRC) for Vehicle to Vehicle (V2V) communication. This algorithm was implemented as an addition to Auburn University’s CACC system and was able to successfully complete automated merging scenarios. The merging scenarios in the front, middle, and back of the platoon were completed at the NCAT test track.

A Cyber-Physical System (CPS) is defined by its unique interactions between digital (cyber) computation and physical motion. Their hybrid nature introduces new attack vectors, but also provides an opportunity to design new security defenses. In this paper, we present a new domain-specific security mechanism, YOLO, that leverages physical properties such as inertia to improve security. YOLO is simple to describe. It goes through two operations: Reset and Diversify, as frequently as possible – typically in the order of a few seconds. Resets mitigate attacks that aim to achieve persistence and enhance the power of diversification techniques. Due to inertia, CPSs can remain safe even under frequent resets. We introduce an analytical approach to evaluate the feasibility of a YOLO-ized system. Using this analytical model we define the constraints on reset periods in order to maintain the CPS’s stability. We evaluate our approach in simulation and on two real systems: an engine control unit (ECU) of a car and a flight controller (FC) of a quadcopter. From our experiments, we determine that resets can be triggered frequently, as fast as every 125ms for the ECU and every second for the FC, without violating safety.