275003 research outputs found
Sort by
Ray-Tracing-Based Micro-Doppler Simulation for 77 GHz Automotive Scenarios
No full text281284Automotive radar sensors can measure Doppler and micro-Doppler signatures caused by road users and their moving parts. Radar sensor validation as well as the development of micro-Doppler-based classification approaches require simulation methods able to generate such signatures. In this paper, a deterministic analytic ray-Tracing approach is introduced which is able to simulate micro-Doppler signatures. The approach is demonstrated with two driving scenarios: A vehicle with rotating tires driving along a straight line and the same vehicle driving in a circle. Based on these scenarios, it is demonstrated that an artefact-free simulation of micro-Doppler can be achieved using the aforementioned simulation method
Electricity Demand and Carbon Footprint of ICT in Germany until 2033 Strombedarf und Carbon Footprint der IKT in Deutschland bis 2033
No full text856863The first results of a new study on the electricity demand and carbon footprint of ICT in Germany from 2013 onwards and with a forecast up to the year 2033 will be presented. This study is part of the joint project Green ICT @ FMD, a BMBF-funded competence centre for resource-conscious ICT. The study provides a view of all ICT in Germany and differentiates between the energy consumption of the usage phase, the energy used in production and the climate-relevant emissions (carbon footprint). The trend forecast distinguishes the main product categories of ICT, referencing the expanded definition that includes connected devices that would not previously have counted as ICT (e.g. televisions, billboards, smart home). The first model shows that the ICT-related carbon footprint is increasing, the use phase continues to be dominant at around 65% and the focus of the environmental load is shifting to the ICT infrastructure, i.e. in particular communication networks and servers or data centres
Is the Future of AI Really Federated? Federated Learning as an Emerging Market
No full text158163The paradigm of centralized training of machine learning (ML) models has come under increasing criticism in recent times. Federated learning (FL) offers a technological extension or even alternative that allows AI models to be trained in a decentralized manner while maintaining data protection requirements and the confidentiality of company data. This opens up a new data space for training specialized, domain-specific ML models in particular. And there are already voices saying that the future of AI is federated. However, it is unclear to what extent FL will actually develop into a real market. In this overview article we analyze the uncertainties that currently characterize FL in a systematic way distinguishing four areas of uncertainty: technology, users & use cases, privacy regulation & IT-security, and suppliers & the FL ecosystem. We conclude that there are clear indications that FL is on its way to become an emerging market. Especially in the application fields of healthcare, banking and manufacturing, FL can solve problems that other privacy-enhancing technologies (PETs) are currently unable to solve. On the other hand, further research is needed to ultimately turn FL into an turnkey application that is easy to deploy for industrial end customers and service providers. Big tech companies could give FL an additional boost in the future by actively embracing the trend towards domain-specific fine-tuning of AI models
Automated image quality assessment for selecting among multiple magnetic resonance image acquisitions in the German National Cohort study
No full textIn magnetic resonance imaging (MRI), the perception of substandard image quality may prompt repetition of the respective image acquisition protocol. Subsequently selecting the preferred high-quality image data from a series of acquisitions can be challenging. An automated workflow may facilitate and improve this selection. We therefore aimed to investigate the applicability of an automated image quality assessment for the prediction of the subjectively preferred image acquisition. Our analysis included data from 11,347 participants with whole-body MRI examinations performed as part of the ongoing prospective multi-center German National Cohort (NAKO) study. Trained radiologic technologists repeated any of the twelve examination protocols due to induced setup errors and/or subjectively unsatisfactory image quality and chose a preferred acquisition from the resultant series. Up to 11 quantitative image quality parameters were automatically derived from all acquisitions. Regularized regression and standard estimates of diagnostic accuracy were calculated. Controlling for setup variations in 2342 series of two or more acquisitions, technologists preferred the repetition over the initial acquisition in 1116 of 1396 series in which the initial setup was retained (79.9%, range across protocols: 73–100%). Image quality parameters then commonly showed statistically significant differences between chosen and discarded acquisitions. In regularized regression across all protocols, ‘structured noise maximum’ was the strongest predictor for the technologists’ choice, followed by ‘N/2 ghosting average’. Combinations of the automatically derived parameters provided an area under the ROC curve between 0.51 and 0.74 for the prediction of the technologists’ choice. It is concluded that automated image quality assessment can, despite considerable performance differences between protocols and anatomical regions, contribute substantially to identifying the subjective preference in a series of MRI acquisitions and thus provide effective decision support to readers.13
Digital-supported problem solving for shopfloor steering using case-based reasoning and Bayesian networks
No full text140145Uncertainty and incompleteness of data challenge the design of knowledge systems for problem solving in shopfloor management. The paper proposes a data-driven design that incorporates traditional means of quality management and goals of production planning and control. It integrates data of a failure mode effects analysis (FMEA) and an 8D problem-solving process into a Bayesian network (BN) embedded case-based reasoning (CBR) cycle. Reducing inconsistencies within the BN, an optimization method uses scoring schemes and structural equation modeling for learning its structure. The results suggest that the optimized BN-CBR system outperforms the single use of CBR in terms of accuracy
Efficient Co-Design Methodology combinig Fast and Accurate System-level Simulations with Transistor-level Characterization
No full textThe Charge-Pump PLL (CP-PLL) stays one of the most popular circuit topologies used for frequency synthesis. Because of its mixed-signal nature, no established linear theory allows an exact description of its behavior, making its design challenging. In this paper an efficient design approach of the CP-PLL is presented achieving accuracy and time efficiency while considering nonlinear and non-ideal effects. A very efficient event-driven simulation model is combined with a CMOS-based design approach. The extracted macroscopic parameters are used at the behavioral level in order to allow a very fast but accurate closed-loop simulation, leading to a robust top-down and bottom-up co-design methodology
Safe and Secure: Mutually Supporting Safety and Security Analyses with Model-Based Suggestions
No full text172181Failures in cyber-physical systems, such as trains and cars, are caused either by faults or attacks. The former are addressed by safety engineering, the latter by security analysis. Both disciplines use separate terminology, processes, and tools. However, both rely on a common system architecture and use models such as component fault trees and attack trees, respectively, for their analyses. We posit that the two disciplines should be aligned without entangling their processes or teams, mutually supporting their considerations. For that purpose, assuming a joint system model, we introduce tool support that heuristically suggests correspondences between analysis elements of the two disciplines and, upon user confirmation, derives additional suggestions for analysis. Our tool allows both disciplines to benefit from the analyses of the other, increasing consistency, exhaustiveness, and alignment of the disciplines. Our paper introduces the approach, describes our prototypical tool, and illustrates the concept with a realistic automotive use case
An Approach to Abstract Multi-stage Cyberattack Data Generation for ML-Based IDS in Smart Grids
No full textPower grids are becoming more digitized, resulting in new opportunities for the grid operation but also new chal-lenges, such as new threats from the cyber-domain. To address these challenges, cybersecurity solutions are being considered in the form of preventive, detective, and reactive measures. Machine learning-based intrusion detection systems are used as part of detection efforts to detect and defend against cyberattacks. However, training and testing data for these systems are often not available or suitable for use in machine learning models for detecting multi-stage cyberattacks in smart grids. In this paper, we propose a method to generate synthetic data using a graph-based approach for training machine learning models in smart grids. We use an abstract form of multi-stage cyberattacks defined via graph formulations and simulate the propagation behavior of attacks in the network. Within the selected scenarios, we observed promising results, but a larger number of scenarios need to be studied to draw a more informed conclusion about the suitability of synthesized data
Chemical tuning of a honeycomb magnet through a critical point
No full textBaCo2(AsO4)2 (BCAO) has seen extensive study since its initial identification as a proximate Kitaev quantum spin liquid candidate. Thought to be described by the highly anisotropic XXZ-J1-J3 model, the ease with which magnetic order is suppressed in the system indicates proximity to a spin liquid phase. Upon chemical tuning via partial arsenic substitution with vanadium, we show an initial suppression of long-range incommensurate order in the BCAO system to T≈3.0 K, followed by increased spin freezing at higher substitution levels. Between these two regions, at around 10% substitution, the system is shown to pass through a critical point where the competing J1/J3 exchange interactions become more balanced, producing a more complex magnetic ground state, likely stabilized by quantum fluctuations. This state shows how slight compositional change in magnetically frustrated systems may be leveraged to tune ground state degeneracies and potentially realize a quantum spin liquid state.108
Gas permeable protection caps for wafer level chip scale packaging (WLCSP) of MEMS environmental sensors
No full text2124This work presents new porous environmental protection caps designed for wafer level chip scale packaging (WLCSP) of MEMS environmental sensors. The caps consist of gaspermeable microstructures formed from loose aluminum oxide powder solidified by a ceramic thin film, grown using atomic layer deposition (ALD). For the first time, the full process flow of the proposed approach is demonstrated by manufacture of the cap wafer followed by substrate bonding using glass-frit technology. By analyzing the influence of the caps on the response time of MEMS humidity sensors, this study proves the viability of the proposed packaging technology. In addition, the ability to further functionalize the porous caps is demonstrated by the deposition of a superhydrophobic polymer thin film by chemical vapor deposition