Projects

RADIANT

  • Radio Astronomy with Low-Complexity Sensors (2019 - present)

    A new low-frequency radio telescope is just about to start operation and challenge our perspective on the universe and the fundamental laws of physics. With more than 100,000 antennas, the low-frequency receiver of the Square Kilometre Array (SKA-low) forms a technology milestone in radio astronomy and, together with its predecessor the Low Frequency Array (LOFAR) featuring approximately 5,000 antennas, constitutes the completion of a paradigm shift towards an all-electric telescope system design. Instead of using a few large parabolic antennas, with the help of supercomputers the SKA-low and LOFAR combine the signals from a massive number of small sensors and synthesize giant low-frequency radio telescopes by smart digital algorithms. While this allows astronomers to perform surveys at unparalleled sensitivity and speed, massive digital sensing as realized today in low-frequency radio astronomy forms an engineering challenge. Due to the enormous number of sensors, the SKA-low produces more data than the worldwide internet, requiring huge quantities of optical fiber, ultra-large memory, and high-performance processing units. While the technological capabilities regarding digital transmission, storage, and computation have exponentially increased during the last decades, the advances associated with analog radio equipment were moderate. Therefore, today hardware cost and power consumption of radio sensors form the main obstacles for constructing future telescopes combining millions of antennas. The project investigates a potentially game-changing approach. The analog complexity of the radio sensors is reduced to its minimum by allowing highly nonlinear behavior. Innovative hardware-aware statistical signal processing methods compensate the undesired effects in the digital domain and an optimized system design with a maximum number of sensors ensures ultra-high performance. Uniting world-class engineers for radio astronomy systems and leading scientists in array signal processing as well as hardware-aware statistical analysis for an initial study of opportunities, the research endeavor aims at: I) quantitatively understanding the potential of radio telescopes with a large number of low-complexity sensors; II) deriving the required high-performance signal processing and data compression algorithms; III) providing proof-of-concept with real radio telescope data sets; IV) planning the verification with a high-performance low-complexity radio telescope prototype. (Project Institutions: Technische Universiteit Delft, Netherlands Institute for Radio Astronomy (ASTRON); Funding: Deutsche Forschungsgemeinschaft (DFG); Role: Principal Researcher)

CATHLIN

  • A Conservative Likelihood Framework for Nonlinear Signal Processing (2016 - 2018)

    The analysis of nonlinear stochastic systems forms a challenge in various branches of science like physics, biology, and computer science. In particular in engineering, where increasing demands for low-cost, energy-efficient and fast sensing devices are emerging, and systems have to be operated outside linear regimes, nonlinear models have gained attention. Recent results for wireless systems show that the analog measurement equipment can be significantly simplified if one allows for highly nonlinear behavior and compensates the effects by optimized system design and strong statistical algorithms. However, to obtain high-resolution measurement systems under these circumstances, a mathematical framework is required to perform the transition from the output of a nonlinear and noisy physical system to an appropriate parametric probabilistic model. Approximating the system output by a particular exponential family distribution has shown to form a versatile method within the setting of parameter estimation with nonlinear system design. Therefore, the project aims at a better theoretical understanding of model replacement strategies and identifying possible applications in wireless systems, biomedical engineering, and machine learning. (Project Institutions: Vrije Universiteit Brussel, Universität Bayreuth; Funding: European Union, Bundesministerium für Bildung und Forschung; Role: Principal Researcher)

SAMURAI

  • Sensor Array Processing for Multipath and Radio Interference Identification and Suppression (2011 - 2014)

    For safety-critical applications based on global satellite navigation systems (GNSS) not only a highly accurate, but also a highly reliable time-delay estimation of the received signal needs to be ensured. Therefore, it is extremely important to develop and explore powerful innovative methods for highly accurate synchronization of navigation receivers, even under difficult reception conditions. The use of array antennas together with appropriate signal processing methods, such as adaptive digital beamforming or high resolution parameter estimation methods (maximum likelihood parameter estimation) has been proven to be one of the key technologies for an effective and highly reliable suppression of multipath and radio interference. Based on previous work in the project IMPOSANT (Interferenz- und Mehrwegeunterdrückung zur hochgenauen Positionierung mit Gruppenantennen) further investigations and developments are conducted in the project SAMURAI. The work can be summarized in three main activities: development of signal processing, development of implementation concepts and testing under realistic conditions. (Project Institutions: Technische Universität München, Rheinisch-Westfälische Technische Hochschule Aachen, Deutsches Zentrum für Luft- und Raumfahrt; Funding: Bundesministerium für Wirtschaft und Energie; Role: Researcher)