Search

Smart adaptive systems are evolving rapidly. In addition to high-performance artificial intelligence from the cloud, smart edge devices are increasingly developing. They are capable of handling constantly more complex classification or even learning tasks independently. After answering the question of how this knowledge can be learned or used at the edge, it must also be determined how this knowledge can be exchanged with a cloud. We would like to investigate this exchange of knowledge between several Edge Runtime Learning Artificial Noses and a cloud intelligence in further experiments. For this purpose, we describe an improvement of our former approaches for an artificial nose. Especially, we implemented a resource-efficient edge nose with lazy learning algorithms on a microcontroller and enabled learning at runtime. This new nose achieves classification rates of up to 92 %, almost as good as the resource-intensive previous version, and forms the basis for research into the knowledge exchange between different edge and cloud AI devices and the processes involved.

In topological insulator thin films, new states of matter can be generated by hybridizing the massless Dirac states that occur at the topological insulator surface. In this work, we experimentally introduce a platform where this hybridization can be continuously tuned: the PbSnSe topological superlattice. Using magnetooptical transmission spectroscopy on high quality molecular beam epitaxy grown PbSnSe/ PbEuSe superlattices, we demonstrate that the energy gap emerging from this hybridization can be tuned with temperature from 1 to 10 THz. The engineering of topological states is a promising road to achieve photo detection and/or photoemission in the terahertz range.

In this paper, we address the challenges facing the adoption of client puzzles as a means to protect the TCP connection establishment channel from state exhaustion DDoS attacks. We model the problem of selecting the puzzle difficulties as a Stackelberg game with the server as the leader and the clients as the followers and obtain the equilibrium solution for the puzzle difficulty. We then present an implementation of client puzzles inside the TCP stack of the Linux 4.13.0 kernel. We evaluate the performance of our implementation and the obtained solution against a range of attacks through reproducible experiments on the DETER testbed. Our results show that client puzzles are effective at boosting the tolerance of the TCP handshake channel to state exhaustion DDoS attacks by rate limiting malicious attackers while allocating resources for legitimate clients.

Large-scale Internet of Things (IoT) applications, like Smart Cities, are ever changing pieces of software. The paradigm of Entity-Component-Attribute (ECA) based software design is well-suited to keep these applications changeable and maintainable. At the same time, W3C working groups propose the Web as IoT convergence platform. Semantic annotations on the data in RDF help tackle cross-domain interoperability issues. While the generation of RDF data from data storage layers has undergone thorough research, the automatic exposure of application run-time data is to this day incomplete. We for this formalize the Entity-Component-Attribute pattern and present an automated mapping to a structure compliant with the Linked Data Platform W3C standard. This structural mapping is then augmented by application domain specific semantics. The result lifts a software design pattern highly suitable for large-scale IoT applications to Linked Data.

While becoming more and more present in our every day lives, services that operate on users' locations or location trajectories suffer from general fear of misappropriation of the transmitted location data. Several works have investigated of how to cope with this drawback. Respective systems claim location-privacy, i.e. keeping users' locations secret, by employing anonymisation techniques concerning a user's identity, or by obfuscating the transmitted location. These approaches lead to a degrade of quality-of-service and can be vulnerable to de-anonymisation attacks, or allow to learn at least the approximate location of a user. Focusing on the application domain of geo-fencing, we present as remedy a protocol that is based on homomorphic encryption of a user's location. The protocol provably provides full location-privacy by non-exposure of the users' location data, while producing exact geo-fencing results. We provide a detailed definition of the protocol, show its applicability in an actual geo-fencing application, and show that the resulting system fulfills all security properties we see for a location-privacy preserving system., 6 Pages, 3 Figures. Accepted for GIoTS 2018. Errata: We used the term 'Public Key Infrastructure (PKI)' in the wrong way. What we meant was an infrastructure of Public Key Cryptography. Thus the 'Certificate Authority and Evaluation Service' is to be considered the 'Key Authority and Evaluation Service'

Virtual Environments have become a compelling tool for various applications beyond gaming and Virtual Worlds, for example for education, collaborative engineering, or simulation and visualization. In the emerging field of smart environments, like Smart Cities and Smart Factories for industry or agriculture, a digital counterpart of a real world site, driven by hundreds of Internet of Things (IoT) sensors that emit their data in realtime, is a central part of the application. Maintaining such a large-scale virtual environment and keeping it up-to-date with changing requirements set by connected sensors and services is a challenge. For this, we present FiVES, a server framework that is based on an aspect-oriented architecture to create highly maintainable large-scale virtual environments by avoiding crosscutting concerns like tangling or scattering of code between modules in resulting applications. We present a fully functional implementation of our system that builds on an extendable and versatile data model, a flexible plugin mechanism, and a transparent yet efficient synchronization layer that is independent of the modules from which applications are assembled.

A doubly-fed induction generator with two power-electronic converters is considered. A mathematical model is given and the flatness of the model is shown. The freedom in the choice of one component of the flat output is used in order to minimize the power losses in the system. The results are supported by an example.

This paper reports on microwave photonics design and characterization of optically-controlled MMIC oscillator. Microwave photoconductive switches (MPCS) design methodology under continuous, intensity-modulated and pulsed optical control has led to microwave predictions of an oscillating device in GaAs integrated technology, under these optical control formats. Microwave oscillations at a frequency of 5 GHz during 2 ns have been experimentally achieved thanks to device illumination using fs optical pulses at 80 MHz repetition frequency.

The paper provides a comparison between an algebraic parameter identification algorithm for the load of a boost converter and classical asymptotic observers for the same purpose. Two asymptotic observers are presented and an algebraic identification algorithm is derived. Aspects of the implementation in fixed point arithmetic are discussed and experimental results are given which highlight the properties of both approaches.

In recent years, piezoelectric transformer (PT) has attracted attention in many practical applications in power electronics field. But the fact that, PT still has some drawbacks and the most issue is the limitation of transferred power. The parallel connection of PTs is introduced in this paper as a solution to the problem that mentioned above. In the experiment part, a high output voltage DC/DC converter based on parallel connection of two market-available Rosen-type PTs are built in order to verify the capable of improving the output power of PT in parallel operating

Highway automation is a promising approach to cope with increasing road traffic and congestions. An important control subtask of an intelligent vehicle/highway system is automatic steering for lateral vehicle control. Due to system uncertainties and the wide range of operating conditions, state-of-the-art robust control techniques are required. This paper introduces two automatic steering controllers for cars driving under highway conditions. The control design is based on sliding mode control and robust state observation. It is shown that good tracking of a reference path, delineated either continuously or discretely, can be achieved with a minimum effort in sensing and without preview of the road curvature.

This paper introduces a robust corrective tracking controller for rigid-link electrically-driven (RLED) robot manipulators operating under motion constraints. Using models of the robot dynamics an environmental constraints, a dynamic model with decoupled position and force degrees of freedom (DOF) is obtained for the mechanical subsystem with respect to a set of task space coordinates. A corrective tracking control is then formulated for the decoupled manipulator dynamics and the electrical actuator dynamics. The controller is robust with respect to parametric uncertainties and additive bounded disturbances in the RLED dynamic model. The controller requires measurement of the robot link positions, link velocities, and motor armature currents.

This work presents a new method for the development of lateral vehicle guidance concepts utilizing steer-by-wire technology. Through steer-by-wire technology the steering command (feedforward) and feedback can be varied for different driving situations. The different combination of steering feedforward and feedback in the human-machine-interface (HMI) influences the driver's perception as well as the haptic interaction between driver and vehicle. New steering concepts may be developed based on appropriate HMI-properties learned from previous manual control research. This work demonstrates that: 1) at low vehicle speeds, the steering concept of "feedforward-force proportional to yaw-rate, in combination with feedback-position deduced from path curvature" is most suitable; 2) at intermediate and high vehicle speeds, the steering concept of "feedforward-force proportional to lateral acceleration, in combination with feedback-position deduced from yaw-rate" should be employed.

It takes a lot of effort to ignite discharge lamps with high ignition voltages but a high frequency operation after the ignition phase makes it even more complicated. In this case the ignition circuit, usually connected in series with the lamp, cannot be neglected during lamp operation. Designing the ignition circuit separate from the rest of the electronic ballast is not suitable. Several ignition concepts are discussed regarding high frequency operation. DC ignition is the only concept that benefits from an increasing operation frequency but other obstacles appear in case of implementation. Even if the problems due to acoustic resonances have been overcome the ignition of the lamp is another issue that favors a low-frequency operation

Fast response of power electronic systems to events like load steps still is a domain of analogue controllers. Fast reconfigurable logic together with high rate analogue to digital converter (ADC) opens this area for digital control. On-site data processing in FPGA minimizes risk of faulty transmission of analogue measurement data

The CoolMOS transistor in combination with the silicone carbide diode from Infineon allows switching frequencies above 1 MHz. To demonstrate the capability's of these devices two applications are developed. First electronic ballasts, where the commonly used 50 kHz for lightning devices scaled up to 2.5 MHz. Three main functions must be integrated into the electronic ballast, these are preheating and igniting the tube and driving the lamp in steady state. For the research, the steady state power supply is selected for this improvement. Second a DC/DC-converter is build up as flyback converter. Both applications show the specific features, needed passives for MHz-power applications and some topology enhancements for switching relief.

We have developed an optical transmitter component for 10.52 Gbit/s data rate at 1310 nm wavelength, using a ridge wave guide distributed feedback laser in a TO-46 type coaxial package with a receptacle for SC or LC fiber connectors and a flexible printed circuit board for connecting the laser diode to a laser diode driver within a 50 /spl Omega/ impedance system. A monitor photo diode measures the time averaged optical output power of the laser diode, while a thermistor detects the component case temperature in the immediate vicinity of the laser. The component is designed for uncooled operation in 10 GbE fiber optics transceivers according to the XPAK or X2 multi-source agreements.

The number of power converters being controlled by a digital hardware is increasing steadily. Modern microcontrollers' and digital signal processors' (DSP) extended computing capabilities enable the implementation of more complex control schemes. Additionally high power applications demand special handling of signal detection and distribution due to electromagnetic interference (EMI) effects. This paper discusses the parallel architecture of a developed voltage regulator module (VRM) and the controller hardware focusing on control strategies. The implementation of those complex strategies in software and in programmable logic devices (PLD) is described in the following. Finally experimental results are compared with simulation data.

The work describes and investigates nonpermanent (pulsed) excitation of piezoelectric transformers (PT). Background of the investigations was the requirement of a higher peak power for a PT. Driven nonpermanently, a FT can be overloaded up to a certain level. Some technical applications require a high power for a finite time, transferred by a relatively small transformer. An example is the ignition of gas discharge lamps as one task for electronic ballast systems. Some of these lamps have a very high ignition voltage, so that the use of a separate piezoelectric transformer for ignition may be useful.

High frequency DC-DC converters are replacing large thyristor converters in the field of DC power supplies in the medium and higher power range. Depending on the given requirements these solutions provide a DC source of higher quality, dynamics and stability for demanding loads. To extend the output power of DC-DC converters to higher ranges the voltage regulator module (VRM) configuration is the most suitable solution. The paper deals with the development of a four-module 100 kW buck converter, operating as a general purpose power supply. The control concept with special respect to the current-sharing technique is also discussed.

The paper applies statistical analysis to chaotic dc-dc converters. A boost converter under current-mode control using a duty cycle limiter is analyzed. A novel stroboscopic map including this limiter is presented. Properties of the converter's inductor current are expressed via the invariant density of the embedded stroboscopic map. The map governs an impulse process. The power density spectrum of this process is calculated and related to that of the inductor current. Expressions for the inductor current ripple and power density spectrum enable the simultaneous consideration of electromagnetic interference and ripple issues during an early design stage.

This paper introduces a new integrated path control strategy (combined path planning and motion control) for autonomous systems of arbitrary dimension. A harmonic artificial potential field is used to specify admissible trajectories leading around obstacles. Using a sliding mode controller, motion is generated along the gradient lines of the potential field, avoiding collisions with the obstacles. The extremely low computational complexity of the proposed method makes it very suitable for online applications. The strategy is applied to mobile robots moving amidst known obstacles. Only the obstacle closest to the robot is considered at each time instance. The switching of the two respective gradient fields along the equi-distance line between two obstacles leads to the interesting phenomenon of an additional "spatial" sliding surface, which is examined in detail. The algorithm guarantees approaching the goal point continuously, following a reasonably short path. Numerical examples are presented to demonstrate the utility of this new strategy. >

Lateral vehicle control for passenger cars, a vital control subsystem of automated highway systems (AHS), has been studied for several decades. Different reference systems have been examined for detecting the lateral vehicle displacement from the lane center. Implementations of 'look-down' reference systems, however, encountered practical constraints, limiting driving speed to approximately 20 m/s under realistic conditions. This paper presents new perspectives on design directions for lateral vehicle controllers under practical AHS conditions for a variety of reference systems. A number of experimental results support the findings of this study.

The artificial potential field method is frequently employed for obstacle avoidance both of mobile robots and of manipulator arms. This paper extends previous results for planar problems to the general n-dimensional case. The complexity is reduced by projecting the n-dimensional workspace onto a 2D subspace called operation plane. Furthermore, only the closest obstacles are taken into account when designing the artificial potential field. Still, global convergence is achieved. The methodology is illustrated with several numerical examples.

This paper approaches DC-DC power converters topologically and statistically. The approach is able to analyze simultaneously periodic and aperiodic behavior. Local c-bifurcation and qualitative analysis are performed on boost converters. The golden mean's importance is highlighted. An analytical expression for the inductor current ripple is derived using statistical analysis. Switching losses during aperiodic operation are compared with periodic ones. Possible applications for an aperiodic DC-DC power converter are suggested.

This work qualitatively analyzes the dynamics of a current programmed mode boost converter operating in continuous conduction mode. The analysis starts with a two-dimensional (2D) map. It proceeds via center manifold theory to a one-dimensional (1D) map. C-bifurcation analysis is applied to the 1D map. The importance of the golden mean is pointed out. This impacts on the power density spectrum (PDS) calculated by statistical analysis.

To fulfil the demands of the energy provider in case of power friendliness of the technical equipment, some years ago one began to develop replacements of conventional magnetic ballasts for fluorescent lamps. As a result, we now have a large number of more or less different topologies, which represent an interesting application type in the lower power range of the power electronics today. Their advantages are high power factor, low crest factor, an almost sinusoidal waveform of the mains current, for instance. In opposition to that we find a high cost factor in development and manufacturing. One way to reduce the costs is the usage of so called single stage topologies to minimize the number of components. The resulting pro and cons are discussed in this paper. Possible ways of dimensioning are explained on several topologies. Finally the basic possibilities of controlling of such circuits is discussed.

The sensor and actuator limits in real-world control systems imposes unavoidable tradeoffs between performance and the size of the operating region. When linear control design techniques are used, these limits are either neglected or accounted for by lowering the gains of the controller. The former results in severe reduction of the operating region, while the latter yields low performance. These tradeoffs can often be mitigated by using a nonlinear control strategy, in which nonlinear terms are intentionally introduced in order to implement different design objectives in different operating regions. In this paper, nonlinear damping is utilized as a design tool in lateral control of automated vehicles. Design considerations of the vehicle lateral controller include lateral tracking errors, passenger ride quality and reference/sensing system characteristics. It is shown that the intentionally introduced nonlinear damping term in the feedback loop possesses several favorable design features in the lateral control system. Experimental results using a full scale automated vehicle show that passenger comfort can be enhanced for small errors by significantly reducing the original controller gains, while at the same time achieving the same level of maximum lateral tracking errors.

This paper addresses the problem of automatic backward driving of passenger vehicles following a specified path. Additional difficulties associated with backward driving, as compared to forward driving are pointed out, with special attention to the sensor location. A special controller structure, based on a concept of kinematic linkage is proposed to overcome the dynamic difficulties associated with backward driving. Simulations and experiments demonstrate its performance and functionality.

In this paper, steering control for passenger cars on automated highways is analyzed, concentrating on look-down reference systems. Extension of earlier experimental results for low speed to highway speed is shown to be nontrivial. The limitations of pure output-feedback of lateral vehicle displacement from the road reference are examined under practical constraints and performance requirements such as robustness, maximum lateral error and passenger comfort. The in-depth system analysis directly leads to a new alternative design direction which allows to harness the benefits of look-ahead reference systems for look-down highway speed automatic driving.