Projet de thèse en Traitement du signal et des images
Sous la direction de Dominique Lesselier et de Mohammed Serhir.
Thèses en préparation à Paris Saclay , dans le cadre de Sciences et Technologies de l'Information et de la Communication , en partenariat avec L2S - Laboratoire des signaux et systèmes (laboratoire) , Signaux (equipe de recherche) et de Université Paris-Sud (établissement de préparation de la thèse) depuis le 01-10-2017 .
Resonant micro-structured 3-dimensional antenna systems, imaging and super-resolution
Background: To improve wireless performance is an issue which is the focus of much attention nowadays, Within that context, at least in a loose sense since the proposed work is set quite upstream with respect to practical challenges, so-called micro-structured systems of wire antennas the inter-element distance of which being much smaller than the electromagnetic wavelengths both/either in transmission and/or reception are to be studied. Context: How to prove and characterize the effective super-resolution or, much better said, the super-localization of a particular radiating element of a micro-structure as sketched above is the main goal of the investigation. The idea behind it is that one should reach a better resolution with respect to the wavelengths of the radiation from a far-field observation followed by some proper back-propagation onto the relevant domain within which the sources and per extension the whole structure are located (with no or little information on the radiating element, nature and localization as well). This is linked with time-reversal concepts already considered in (scalar) acoustics and further explained in the applied mathematics community via the example of 2-D Helmholtz resonators, while, in parallel, again in a similar 2-D scalar situation, resolution enhancement involving homogenization can be exhibited. Yet, results in the vector-wave 3-D situation of our interest are preliminary, and to the best of our knowledge absent from the academic literature. So, significant challenges must be overcome, requiring extensive expertise and sustained effort: (i) in computational fashion ¬ we are interested in partially periodic structures of limited extent, open or partially open (above a metal plate), in particular with facing vector wave field equations so as to get, in best circumstances, Green dyads in complex pseudo-cavity regime; (ii) in more practical terms we must take into account the realization constraints of set-ups, the need of wideband measurements or transient measurement systems, and correspondingly geometrical and electrical uncertainties that theories as built may not account for in the proper way. (iii) in terms of explanation/understanding of observed results which remain quite controversial in the communities of antennas and inversion, e.g., the notion of coding far fields by resonances of pseudo-cavities as set forth by applied mathematicians. Success of above endeavors relies in particular on availability of electromagnetic 3-D modeling tools of such micro-structured structures at emission/reception. Commercial codes (say, CST Microwave Studio, HFSS, etc.) can be employed to simulate but they are subject to heavy implementation and this one must be very neat. Yet the observed resonances, tracked to the behavior of dyads Green characteristics of impulse responses of structures under analysis, impose to take into account relevant power feeds and/or realized set-ups. Also, the asymptotic analysis of these dyads, mainly for explaining super-resolution, remains an open issue, noticing that full transient data might end up in time-harmonic frequency-diverse data, and carefully tailored back-propagation algorithms might substitute to traditional TR mirrors. We in some way need to extend the analysis of scalar 2-D acoustic cases (the Helmholtz resonators aforementioned) to vector 3-D cases of interest, at least to provide a satisfactory interpretation even at the price of model simplification. We also aim at revisiting theories of singularities analysis originally developed for characterizations of remote targets. Overall we have to consider what is meant by super-resolution/localization, since information is carried by evanescent waves collected in near-field yet also transported into mode-structured far-field easier to measure, spatial/temporal focusing in such a multi-path environment being expected. Work plan: (The Chinese cursus of the PhD candidate is mostly to impact the 1st year.) WP1 (months 1-12) Learning about state of art, definition of theoretical/numerical/ experimental steps: to understand state-of-art at direct modeling/ inversion/imaging level, + gain grasp of existing solution tools/experimental challenges. WP2 (months 6-24) Modeling, numerical analysis, choice of scenarios: to develop modeling algorithms of electromagnetic behaviors in situations of interest with changes (geometry, material parameters, feeds, etc.) of structures, while to involve with laboratory-controlled measurements, + to propose imaging relevant to time-reversal at large. WP3 (months 12-36) Development and testing: to build a software demonstrator and test direct modeling and imaging solutions, with confrontation to experimental laboratory-controlled data available. WP4 (months 18-36) Dissemination: to disseminate via contributions at symposia & journals. 1st conference contributions to be proposed/given in 2nd year, writing articles started after 24th month, most work last 6 months. Available expertise & cooperation: The work is to be run within a co-operation between the L2S laboratory (mathematical analysis, construction of models, numerical simulations, mostly using in-house codes to be developed) and the GeePs laboratory (numerical simulations as well, mostly using commercial codes, controlled-laboratory experiments on properly down-scaled set-ups in anechoic chambers). In that framework the PhD student will be advised from L2S, within the STIC Paris-Saclay Doctoral School, by D. Lesselier, in view of his broad upstream expertise. A close-knit cooperation will be conducted with M. Serhir, Associate-Professor of CentraleSupélec and at GeePs, due to his fine knowledge of numerical simulations and experiments in the microwave domain. The broad expertise of L2S in signal processing, modeling and inversion is proven, as the one of GeePs for whatever concerns the science of antennas. The PhD investigation is to be fruitfully discussed with the Department of Mathematics, ETH Zurich (H. Ammari and team) and Department of Mathematics, Hong Kong University of Science and Technology (H. Zhang), in view of the sophisticated analysis behind it. Linkage could be set up with Chinese partners (it exists already with University of Electronic Science and Technology of China, Chengdu, and is being developed with Beihang University also, others can/should be pursued).