Using tensor contractions to derive the structure of slice-wise multiplications of tensors with applications to space-time Khatri-Rao coding for MIMO-OFDM systems. - In: EURASIP journal on advances in signal processing, ISSN 1687-6180, Bd. 2022 (2022), 109, S. 1-26
The slice-wise multiplication of two tensors is required in a variety of tensor decompositions (including PARAFAC2 and PARATUCK2) and is encountered in many applications, including the analysis of multidimensional biomedical data (EEG, MEG, etc.) or multi-carrier multiple-input multiple-output (MIMO) systems. In this paper, we propose a new tensor representation that is not based on a slice-wise (matrix) description, but can be represented by a double contraction of two tensors. Such a double contraction of two tensors can be efficiently calculated via generalized unfoldings. It leads to new tensor models of the investigated system that do not depend on the chosen unfolding (in contrast to matrix models) and reveal the tensor structure of the data model, such that all possible unfoldings can be seen at the same time. As an example, we apply this new concept to the design of new receivers for multi-carrier MIMO systems in wireless communications. In particular, we consider MIMO-orthogonal frequency division multiplexing (OFDM) systems with and without Khatri-Rao coding. The proposed receivers exploit the channel correlation between adjacent subcarriers, require the same amount of training symbols as traditional OFDM techniques, but have an improved performance in terms of the symbol error rate. Furthermore, we show that the spectral efficiency of the Khatri-Rao-coded MIMO-OFDM can be increased by introducing cross-coding such that the “coding matrix” also contains useful information symbols. Considering this transmission technique, we derive a tensor model and two types of receivers for cross-coded MIMO-OFDM systems using the double contraction of two tensors.
https://doi.org/10.1186/s13634-022-00937-5
Empirical estimation of ETSI ITS-G5 performance over an IPv6-based platform. - In: 2022 IEEE International Performance, Computing, and Communications Conference (IPCCC), (2022), S. 185-193
Intelligent transport systems (ITS) promise to leverage wireless communications among vehicles. Performance of wireless communication is of crucial importance and can severely impact safety, efficiency and infotainment of transport applications. In this paper, in order to promote the usage of IPv6 in vehicular networks, we deploy a wireless communication framework in a Linux environment based on the Internet protocol (IP). Further, we evaluate the behaviour of the ITS-G5 in various stationary and mobile scenarios to investigate the impact of environmental and radio factors on the performance of ITS-G5. Several measurements such as inter-reception time (IRT), packet loss, latency, end-to-end (E2E) latency, reception position error (RPE), and throughput are carried out to discover the limitations and strength of the technology in a real world test-bed.
https://doi.org/10.1109/IPCCC55026.2022.9894344
Deep learning aided interpolation of spatio-temporal nonstationary data. - In: 30th European Signal Processing Conference (EUSIPCO 2022), (2022), S. 2221-2225
Despite the growing interest in many fields, spatio-temporal (ST) interpolation remains challenging. Given ST nonstationary data distributed sparsely and irregularly over space, our objective is to obtain an equidistant representation of the region of interest (ROI). For this reason, an equidistant grid is defined within the ROI, where the available time series data are arranged, and the time series of the unobserved points are interpolated. Aiming to maintain the interpretability of the whole process while offering flexibility and fast execution, this work presents a ST interpolation frame-work which combines a statistical technique with deep learning. Our framework is generic and not confined to a specific application, which also provides the prediction confidence. To evaluate its validity, this framework is applied to ultrasound nondestructive testing (UT) data as an example. After the training with synthetic UT data sets, our framework is shown to yield accurate predictions when applied to measured UT data.
https://ieeexplore.ieee.org/document/9909600
Estimating multi-modal dense multipath components using auto-encoders. - In: 30th European Signal Processing Conference (EUSIPCO 2022), (2022), S. 1716-1720
We present a maximum-likelihood estimation algorithm for radio channel measurements exhibiting a mixture of independent Dense Multipath Components. The novelty of our approach is in the algorithms initialization using a deep learning architecture. Currently, available approaches can only deal with scenarios where a single mode is present. However, in measurements, two or more modes are often observed. This much more challenging multi-modal setting bears two important questions: How many modes are there, and how can we estimate those? To this end, we propose a Neural Net-architecture that can reliably estimate the number of modes present in the data and also provide an initial assessment of their shape. These predictions are used to initialize for gradient- and model-based optimization algorithm to further refine the estimates. We demonstrate numerically how the presented architecture performs on measurement data and analytically study its influence on the estimation of specular paths in a setting where the single-modal approach fails.
https://ieeexplore.ieee.org/document/9909796
Time-domain analysis of ultra-wideband scattering properties of fruits. - In: 2022 19th European Radar Conference, (2022), S. 77-80
In the present paper we evaluate scattering properties of fruits measured with a short-range Ultra-Wideband radar. This is part of our investigation how effectively such a radar can be used to infer information such as fruit biomass or ripeness in an agricultural environment. The covered frequency band spans from 1.4 to 5.6 GHz. We analyze measured impulse responses of a watermelon, a grapefruit, and an apple with respect to a dependency on the distance between radar and fruit and the observation angle i.e., rotation of the fruit. Measurements are performed under laboratory conditions, however, we analyze the data considering a pre-harvest analysis on a field. It becomes apparent that an analysis of the dispersed dominant reflection of the peel is most promising. Due to the natural growth and hence anisotropy of the fruits, we conclude to average over multiple monostatic observation angles to reduce the natural variations of e.g. the scattered power.
https://doi.org/10.23919/EuRAD54643.2022.9924720
Link budget and design approach of a non-terrestrial 5G automotive antenna. - In: 2022 52st European Microwave Conference, (2022), S. 864-867
5G low-earth orbiting satellites are continuously increasing attention from automotive industry for automated and connected driving. Compactness of user equipment antennas and high data rates are key performance figures for efficient satellite communication systems. Here, we present a link budget for internet-of-things applications at Ka-band frequencies (5G frequency range FR2). Anticipating a realistic high-gain satellite antenna, an uplink data rate of 4 Mbit/s can be achieved with a compact user terminal antenna with a moderate gain of 13 dBi. Along these lines, a 4×4 patch antenna array was designed for seamless embedding in the plastic part of a car body, in order to verify the link budget calculations by experiment. The radiation performance was measured under free-space conditions and with the antenna embedded in the rear spoiler wing of a modern passenger car. The array offered 11.2 dBi realized gain and 1.6 GHz of −10 dB matching bandwidth, with an uplink data rate of 2 Mbit/s, promising for many mobility applications.
https://doi.org/10.23919/EuMC54642.2022.9924379
Evaluation of scenario-based automotive radar testing in virtual environment using real driving data. - In: 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC), (2022), S. 2379-2384
Safety assurance of intended functionality through rigorous testing is a key to large-scale homologation and deployment of automated driving. It is therefore imperative to transfer real world tests into efficient and quantifiable virtual testing procedures and environments without compromising reliability. In earlier work, we presented a fully operational over-the-air vehicle-in-the-loop test system for automotive millimeter-wave radar, where we generated a virtual electromagnetic environment with physically realistic radar target echoes. We evaluated the performance of the implemented test system with an exemplary scenario parameterised with analytically pre-defined vehicle manoeuvres. In this work, we significantly proceed with the performance evaluation through re-simulation of scenarios based on real driving data and traffic manoeuvres. We have measured a standard Euro-NCAP scenario, namely, the Car-to-Pedestrian Longitudinal Adult on a proving ground and re-simulated the ground truth parameters in the test bed. We compare the consistency of the test results at several data abstraction levels using parameter trajectories. Additionally, we introduce and evaluate quality metrics such as difference and root mean square error. For a driving scenario approximately 20 seconds long, we achieved promisingly low root mean square errors in range, azimuth and RCS of 0.3 m, 0.5˚ and 2 dB, respectively.
https://doi.org/10.1109/ITSC55140.2022.9922366
Human RF electromagnetic exposure to V2X-communication. - In: Advances in radio science, ISSN 1684-9973, Bd. 19 (2022), S. 233-239
In the era of automated and connected driving, more and more cars will be equipped with wireless transmission technologies such as mobile communications 4G (LTE) and 5G, WiFi, Bluetooth, and V2X. For the technical implementation of V2X-communications, different standards like cellular-V2X from the cooperation 3rd Generation Partnership Project and ITS-G5, based on the WiFi standard 802.11p from the Institute of Electrical and Electronics Engineers, are under consideration. The electromagnetic environment of cars and the corresponding exposure of the general public to wireless emission will be significantly influenced by new radio technologies. Under all circumstances, it must be ensured that the exposure of the electromagnetic fields inside a car does not cause any harmful effects on humans. In order to quantitatively assess the resulting exposure, the generated exposure must be correctly recorded and evaluated according to their specific time-frequency spectra. This paper describes a new measurement procedure suitable for the V2X-standard ITS-G5 together with various exposure measurements performed in different cars with WiFi, Bluetooth and ITS-G5. In comparison of all wireless standards studied here, the V2X-service generated the highest electric field strengths inside a car, when a transmitting di-patch antenna was mounted on the windscreen inside the driver's cabin. The maximum fraction of the corresponding ICNIRP reference level amounted to 15.1 %. We conclude that the total exposure of wireless on-board automotive devices will be dominated by ITS-G5, if the transmitting antenna is located inside the passenger cabin. As V2X-communications will increasingly penetrate road traffic, the resulting exposure should be carefully monitored, in order not to exceed the corresponding reference levels for general public.
https://doi.org/10.5194/ars-19-233-2022
Tensor-Based Machine Learning Approaches for Financial Time-Series Prediction. - Ilmenau. - 68 Seiten
Technische Universität Ilmenau, Masterarbeit 2022
Traditionelle und tensorbasierte Methoden des maschinellen Lernens sind in Anwendungen wie der Finanzsignalverarbeitung, der Computer Vision, der Kommunikationstechnik und vielen anderen Bereichen weit verbreitet. In dieser Arbeit vergleichen wir die Leistung traditioneller Ansätze des maschinellen Lernens mit den tensorbasierten Ansätzen bei der Vorhersage von Finanzzeitreihen. Dabei formulieren wir das Problem sowohl als Klassifikations- als auch als Regressionsproblem. Bei der Klassifizierung wählen wir als Ziel das Vorzeichen der zukunftigen prozentualen Veränderung des Schlusskurses, während bei der Regression der zukünftige Schlusskurs das Ziel ist. Zunächst betrachten wir zwei bestehende überwachte tensorbasierte maschinelle Lernalgorithmen: Higher-Rank Tensor Rinde Regression (hrTRR) und Least-Squares Support Tensor Machine (LS-STM). Darüber hinaus verwenden wir als Vorverarbeitungsschritt einen bestehenden unüberwachten Tensor-basierten Ansatz zur Merkmalsextraktion, der auf der globalen Tucker-Repräsentation der Trainingstensoren basiert. Unseres Wissens nach ist dies die erste Arbeit, die die globale Tucker-Darstellung und hrTRR im Kontext von Finanzzeitreihen anwendet. Darüber hinaus erweitern wir die tensorbasierte Zeitreihendarstellung aus einer früheren Arbeit auf den allgemeinen Fall, so dass wir den Beobachtungshorizont sowie die Anzahl der zeitversetzen Tensor-Slices innerhalb des Modells frei wählen können. Um die besten Werte für den Beobachtungshorizont und die Anzahl der zeitverschobenen Tensor-Slices zu finden, wenden wir eine Rastersuche auf einer Validierungsmenge an und wählen die Werte aus, die den niedrigsten mittleren absoluten Fehler in Prozent ergeben. Darüber hinaus wird ein Ansatz, der die Vorhersageperiode für eine unterschiedliche Auswahl des Beobachtungshorizonts und der Anzahl zeitlich verschobener Slices vereinheitlicht vorgestellt. Schließlich bewerten wir die Algorithmen anhand von OHLCV-Kryptowährungs-daten für verschiedene Zeitrahmen. Anschließend betrachten wir einen Kreuzvalidierungsansatz mit gleitendem Fenster, um die Modelle über mehrere Batches zu evaluieren, und ein anpassbares gleitendes Fenster, das den Einfluss der Anzahl der Trainingsmuster pro Batch auf die Vorhersageleistung bemisst. Die Ergebnisse zeigen, dass die tensorbasierten Ansätze besser sind gegenuber ihrem vektorbasierten Gegenstück. Außerdem verbessert die Verwendung eines längeren Beobachtungshorizonts die Vorhersageleistung für alle Experimente, wenn auch in unterschiedlichem Ausmaß. Durch die Anwendung der Tensor-Entrauschung stellen wir außerdem fest, dass die Modelle einen größeren Beobachtungshorizont nutzen.
Quality of service in software-defined wireless sensor networks. - Ilmenau. - 93 Seiten
Technische Universität Ilmenau, Masterarbeit 2022
Die Nachfrage nach einer effektiven Dienstqualität wurde durch die rasche Ausweitung von rasante Ausbreitung von drahtlosen Sensornetzanwendungen. Als drahtlose Sensornetzwerke Sensornetzwerke entwickelt wurden, stand der Energieverbrauch im Vordergrund, während die Anforderungen an die Dienstqualität weniger wichtig waren. Da jedoch Multimedia-, kritische und Echtzeitanwendungen zunehmen, ist es wichtiger denn je, qualitativ hochwertige Dienste anzubieten. Dienste. Die effektive Bereitstellung der erforderlichen Dienstqualität ist eine große Herausforderung, da drahtlose Sensoren nur über begrenzte Ressourcen verfügen. Die Schwierigkeiten, denen sich WSNs konfrontiert sind, können durch den Einsatz von Software Defined Networking als Routing-Methode gelöst werden. Dieses Projekt schlägt eine Methode vor, bei der ein Controller eine Ansicht des Netzwerks erstellt und den Dijkstra-Algorithmus verwendet, um Pakete über den kürzesten Weg zu leiten. Die Architektur besteht aus aus Gateways, die Daten von den drahtlosen Sensorknoten empfangen und dann die Daten an den Senkenknoten weiterleiten, basierend auf den Anweisungen des Controllers. Wenn dieser Ansatz im Gegensatz zum konventionellen Routing-Protokoll für leistungsarme und verlustbehaftete Netze eingesetzt und verlustbehaftete Netze eingesetzt wurde, zeigen die Ergebnisse unserer Simulation eine deutliche eine deutliche Verbesserung der Dienstgüte des Netzes.