Search

As driving task has been and still is the domain of humans, they are a relatively good model of how to behave on road. However, as we would like to constantly raise standards of safety and efficiency of transportation by automated driving, simple imitation of human driving seems not to be good enough. In this paper, we present a method for learning a baseline agent represented as a neural network, which is responsible for planning the behaviour of an automated car, based on a set of expert demonstrations. We show that imitation learning could be a promising way for achieving initial policy, which is almost as effective as human. The generated policy could be then fine-tuned by the reinforcement learning methods. We show that such an approach suits well for an automated driving task and allows faster training of the RL-based artificial driver.

Future emergency response and disaster recovery operations will increasingly rely on an up-to-date operational picture obtained from IoT devices deployed in the operational theatre. Some of these IoT devices will be present in the environment as a part of the civilian smart infrastructure, while others will be a part of the deployable military infrastructure. We present a high-level design for a federated smart environment enabling secure civil-military information sharing in support of emergency response and disaster recovery operations. Our security architecture, relying on standardized protocols, data formats and interfaces, spans from physical sensors through network and processing layers to applications. The proposed novel security mechanisms cover confidentiality, integrity and availability of information at rest, in transfer and in use.

Modern multi-nation military communication and information systems demand rapid deployment and reconfiguration to enable secure information exchange between domains belonging to different nations. In order to support such a multi-nation mission scenarios NATO has been developing the concept of a federated mission network (FMN). These networks are leveraging the advantages of software-defined networking (SDN) and network function virtualization (NFV) to adapt to a wide range of security requirements using network security functions (NSF). The investigation reported in this paper discusses two use cases, namely how to automate security policy evaluation, and how to deploy a security guard (Information Exchange Function) between network segments with different classification. Both cases use network scenarios from protected core network (PCN) because the goal is to compile open challenges to automate the deployment and management of secure coalition networks.

This paper presents a simulation and experimental study focused on the operation of the EV chargers based on a four-wire Vienna rectifier and two isolated DC-DC converters. A standard Vienna rectifier switching at low frequency can co-operate with a three-phase grid providing sinusoidal and balanced currents with help of DC-DC converters (“three-phase PFC”). The paper shows that the same system may be immune to grid faults as can continue operation without one of three phases. Moreover, it can be also supplied from the single-phase grid after a slight modification of the topology. All these operation modes are carefully analyzed and simulated in Saber. Finally, initial experiments of the laboratory model of 5 kVA charger are also shown to prove correct operation in three-and single-phase modes.

This work presents a silicon integrated high-performance thin-film inductor technology with improved magnetic core laminations. The deposition process parameters of the lamination material are tuned to obtain good quality lamination oxide, and its quality is verified by analyzing elemental composition through energy-dispersive X-ray (EDX) spectroscopy. Due to the high-quality lamination oxide, a peak quality factor of 21 and inductance to DC resistance (L/R DC ) ratio of 425 nH/Ohm are achieved in this inductor technology. These figure of merits represent more than 95% inductor switching efficiency in power converters, and significant improvements in quality factor (Q) and L/R DC compared to other leading technologies.

This paper presents a medium voltage (1.5 kV) flying capacitor DC/DC converter rated at 30 kW. The fast switching 1.2 kV/11 mΩ SiC MOSFETs converter operates in quasi-2-level (Q2L) mode at high frequency (100 kHz) to obtain a low size of the flying capacitor (330 nF). Moreover, a novel switching pattern is introduced and ZVS at turn-on is also enabled. Such control method leads to lower switching losses, higher switching frequency and minimized inductor volume compared to more conventional topologies. Presented experimental results prove correct operation of the proposed converter and control method.

This paper is focused on an alternative concept of the EV off-board charging system based on three-level Vienna rectifier operating at grid frequency and two soft-switched, isolated current-source DC-DC converters. Operation principles of the system including a low frequency input stage and DC-DC converters are studied, also using circuit simulations. Then, design issues of vital parts are presented. Finally, a 5kVA laboratory model is shown and selected measurements are provided and discussed.

This paper presents various issues related to a medium voltage power switch designed with series-connected transistors of the 1700 V/300 A SiC MOSFET power module. The focus of the paper are the voltage imbalance compensation methods: passive snubber circuits and active gate driving. Both methods are simulated in Saber software with the use of detailed models of the transistors and diodes of the module. At first, elements of the DRC snubber were determined and simulated during the double pulse procedure. Then, new active gate driver is proposed and verified via series of simulations. Finally, the test setup was designed with two series connected SiC MOSFETs inside CAS300M17BM2 power modules and experimental tests were performed at 1.5 kV DC to compare two compensation methods.

In this paper we introduce a novel risk-aware routing approach for software defined networks vulnerable to cyber attacks. We focus on continuous routing supervision with an objective of minimizing operational risk faced by the users due to utilization of imperfectly secure network elements, such as links and switches, for transmission of data with different protection requirements. Initial tests confirm that the proposed routing mechanism, adapting to changes in the level of threat to which network elements are exposed, can help to increase the customers’ satisfaction, expressed by the volume of successfully completed service requests, while ensuring an efficient use of network resources and maintaining the acceptable level of risk associated with data transfer.

Blockchains offer an interesting solution to some of the security challenges encountered in military applications. They are particularly attractive in the scenarios, where it is difficult or even impossible to establish a common root of trust, e.g., in the context of civil-military collaboration, where military organizations need to build trusted information exchange infrastructure with various types of civilian governmental and non-governmental organizations, local communities, commercial companies and private persons. In our work, we discuss how blockchain can be used to store metadata describing information collected from the IoT devices owned by the federation members as well as crowdsourced from sensors belonging to private users. We present a high level architecture for a metadata binding solution using blockchains compliant with STANAG 4774 and 4778. We also describe a concrete technical solution based on Hyperledger Fabric and some of the open issues.

Natural disasters occur unpredictably and can range in severity from something locally manageable to large scale events that require external intervention. In particular, when large scale disasters occur, they can cause widespread damage and overwhelm the ability of local governments and authorities to respond. In such situations, Civil-Military Cooperation (CIMIC) is essential for a rapid and robust Humanitarian Assistance and Disaster Relief (HADR) operation. These type of operations bring to bear the Command and Control (C2) and Logistics capabilities of the military to rapidly deploy assets to help with the disaster relief activities. Smart Cities and Smart Environments, embedded with IoT, introduce multiple sensing modalities that typically provide wide coverage over the deployed area. Given that the military does not own or control these assets, they are sometimes referred to as gray assets, which are not as trustworthy as blue assets, owned by the military. However, leveraging these gray assets can significantly improve the ability for the military to quickly obtain Situational Awareness (SA) about the disaster and optimize the planning of rescue operations and allocation of resources to achieve the best possible effects. Fusing the information from the civilian IoT sensors with the custom military sensors could help validate and improve trust in the information from the gray assets. The focus of this paper is to further examine this challenge of achieving Civil-Military cooperation for HADR operations by leveraging and fusing information from gray and blue assets.

The RICH detector of the NA62 experiment is one of the key detectors to achieve the muon rejection needed in the search for the ${K^ + } \to {\pi ^ + }\nu \bar \nu $ decay, performed by NA62. Since $BR({K^ + } \to {\mu ^ + }\nu )$ is higher than $BR({K^ + } \to {\pi ^ + }\nu \bar \nu )$ by more than 9 orders of magnitude, it represents one of the most relevant background contributions. Its rejection is performed with kinematic reconstruction of the event and identification of the charged particles in the final state (pions against muons). The Ring Imaging Cherenkov detector (RICH) gives a muon rejection higher than a factor 100 in the momentum range 15-35 GeV/c, and measures the track crossing time with a resolution better than 100 ps. It is also exploited to provide trigger for charged particles. In this contribution the RICH detector will be described, together with its basic performance, the mirrors alignment procedure and its usage in ${K^ + } \to {\pi ^ + }\nu \bar \nu $ search.

This work presents a CMOS back-end-of-line (BEOL) compatible thin-film magnetic inductor technology to realize fully integrated on-chip power conversion. The conventional way of realizing on-chip inductors results in poor inductance density. The utilization of magnetic core helped to boost the inductance. The measured electrical performance confirms that a high inductance density of 140 nH/mm2 can be achieved, which is 30 times more than its air-core counterparts. Due to the inductance boost, large value inductance can be realized without sacrificing its DC resistance, and the inductance to DC resistance ratio reaches up to 320 nH/Ohm. These values translate into more than 95% inductor efficiency, and eventually, it will facilitate to realize fully integrated, high efficiency, Power Supply On-Chip (PwrSoC).

Distributed ledgers and blockchains have been heralded as a new and universally applicable architecture for management of trust. In this article we investigate what are the possible advantages of using distributed ledger technology within the North Atlantic Treaty Organization (NATO) operations and what are possible use cases for such use. We also analyze different types of distributed ledgers proposed in the literature in respect to their applicability in context of federated military operations. Finally, we describe in more detail a specific use case, related to use of blockchain as new approach to implementing trusted binding mechanism for detached labels as stipulated in the NATO Standardization Agreement (STANAG) 4778, which is now undergoing a ratification process within NATO.

The active casing approach is to reduce device noise by controlling the vibration of its casing. Its efficiency was confirmed by the authors in laboratory experiments, and it was reported in previous publications. However, the casing was distant from walls of the enclosure. Now, a new configuration is considered, when the casing is placed at a wall, and the actuators mounted on the casing from the side of the wall are also employed to actively reduce the emitted noise. In such a configuration the acoustic reflection from the wall is intentionally used. The considered wall is smooth and rigid, thus it is highly acoustically reflective. In this paper, a preliminary experimental investigation of such concept is presented. The primary and secondary paths analysis is given. Advantages and limits of the proposed approach are pointed out and discussed.

In this paper the fundamental acoustic camera issues have been presented. First, design and construction of the 2D microphone array with a vision camera have been described. Finally, acoustic camera has been improved by the fusion of the sound map and digital image from the vision camera. The authors describe in details some fundamental issues related to the acoustic camera construction. Experiments have been performed with NI cDAQ acquisition system equipped with $4\times 4$ microphone array including 2D vision camera. The experiments have been concentrated on the evaluation of different image fusion algorithms performance. In result, for the specific purposes, especially for noisy spaces, some alternative method of the image fusion might be more efficient than standard algorithm.

Managing risk related to operation of Communication and Information Systems (CIS) is nowadays a very important objective of every organization. Incorrectly performed, risk assessment may lead to critical errors in the entire risk management process and expose an organization to unacceptable operational risk. Unfortunately, most of the currently adopted risk assessment approaches are only marginally applicable to modern IT systems. In this paper we evaluate suitability of various existing risk analysis approaches in support of risk management process for operation of Software Defined Network (SDN). Our objective is to identify risk analysis methods, which are most suitable for supporting risk-aware routing in SDN. From a large set of available risk analysis methods, we select methods that can be potentially used for risk-aware routing system in SDN and compare their suitability for this task. We conclude that the most suitable for our use case are risk assessment methods based on Bayesian networks.

This paper presents a methodology and mathematical foundation for the calibration of the microlens array position used in light field cameras. The accurate estimation of optical system parameters is crucial for performing a 4D light field reconstruction based on 2D camera intensity images. The method presented in this paper allows the calculation of four distinct parameters which characterize the microlens array position: microlens pitch, rotational offset and translational offsets in two directions. All of the parameters are calculated with relation to the image sensor of a camera. The calibration procedure involves capturing a single image of an evenly illuminated, plain white surface. The center point of each lenslet is assessed using a center-of-gravity method, and the obtained coordinates are then used to calculate the parameter values as a solution for non-linear optimization problems in least squares sense.