In fast changing markets, customers worldwide rely on Thales. Thales is a business where brilliant people from all over the world come together to share ideas and inspire each other. In aerospace, transportation, defence, security and space, our achitects design innovative solutions that make our tomorrow's possible.
Studierichting(en): Elektrotechniek, Toegepaste Wiskunde, Technische Natuurkunde
Titel van de opdracht:
Information-based Processing in Radar and Communications: Compressive Sensing, Information Geometry and Machine/Deep Learning
With 80.000 talents working in 68 countries, we are one of the biggest high-tech employers in the field of safety and security. In the Netherlands, where 2000 employees are based, we are located in four cities: Huizen, Delft, Eindhoven and Hengelo (HQ). Together with an extensive ecosystem of knowledge partners, customers and suppliers, we work on radars for naval vessels, cyber security solutions, transportation systems, communication equipment for land forces, cryogenic cooling solutions, research & development for radar tech (in collaboration with TU Delft).
Compressive Sensing (CS) is a recent paradigm in sensing (since 2004) that works with a reduced number of measurements for a comparable sensing result. It is based on the incoherence of the sensing and sparsity of the processing results. Its major parts are: compressive data acquisition and sparse-signal processing. Most promising benefits of CS in radar are high resolution and multi-target analysis. Other specific issues in applying CS in radar or communications are: coded and sparse sensing (via waveforms and antenna arrays), noise and clutter, grid design and match, real-time implementation, etc.
Information geometry (IG) raises a new approach to stochastic signal processing (since the eighties) as its main principle is that many important structures in the stochastic signal processing can be treated as structures in differential geometry. Most promising benefits of IG have been found in resolution bounds and parameter estimation.
The importance of information in data is stressed in both fields as the useful dimension of signals is much smaller than the data dimensionality. Accordingly, conventional processing can be improved if the demands of data acquisition and signal processing are optimized to the information content (which links it also to the information theory).
IG and (Bayesian) CS can also be used in understanding of machine/deep learning (MDL), especially in the stochastic analysis of the underlying processing layers for better performance. In particular, using information distances can improve the performance.
ABOUT THE ASSIGNMENT
Thales NL proposes an internship project whose aim is to investigate applicability of the information-based processing with emphasis on practical issues in signal processing (SP). Simulated data are to be used to demonstrate the applicability in realistic cases.
Proposed project planning:
Thales Nederland is part of the Thales Group, with 65,000 employees in over 50 countries, including 22,000 working in R&D. This makes Thales one of Europe's largest electronics companies and it has an extremely strong position worldwide. Thales Nederland is active in the sectors of Defence, Security and Transportation Systems. We have almost 2,000 employees, making us the leading supplier...