An overview of our lab and our activities. Giuseppe Vecchi March 2015

Transcription

1 An overview of our lab and our activities Giuseppe Vecchi March 2015

2 ISMB Research Areas Navigation Technologies Pervasive Technologies Multi-Layer Wireless Solutions ISMB ISMB is organized in Research Areas focused on core sectors of ICT. The areas are capable of managing the whole value chain, from basic technology up to its practical implementation (proof-of-concept). Advanced Computing & Electromagnetics Antenna & EMC Lab (LACE) Applied Photonics Mobile Solutions Innovation Development LACE is: a partnership lab with Politecnico di Torino (POLITO) one of the High Quality Laboratories (LAQ) of Politecnico di Torino (strategic research action) 2

3 LACE RESUME ISMB-LACE STAFF 4 RESEARCHERS (80% WITH A PHD) POLITO-LACE STAFF 2 FULL PROFESSORS 1 ASSOCIATE PROFESSOR 2 POST-DOC 5 PHD STUDENTS 3 MAIN RESEARCH LINES 1)COMPUTATIONAL ELECTROMAGNETICS 2)ANTENNA DESIGN 3)BIOMEDICAL IMAGING 3

4 ANTENNA TESTING FACILITIES Shared with Politecnico di Torino: Anechoic chamber for microwave and mm-waves (above 1.5 GHz) Outdoor antenna test range up to 40 GHz Spherical near field test range in anechoic chamber (from 700 MHz to 40 GHz) Instrumentations for measurements up to 90 GHz 4

5 EMC WHO WE ARE Antenna siting: Space, Automotive Analytic-based Computational Techniques RF tomography for lesion detection: breast cancer, brain and bone anomalies Prediction/Simulation: Finite Element Method (FEM) Integral Equations/Boundary Element Method Special Multi-Scale algorithms for highly complex structures Domain Decompositions Fast Factorizations/Solvers Optimization/Synthesis/ Imaging: Genetic, Particle Swarm Multi-objective Alternate Projections Compressed Sensing Linear and non-linear inversion High-performance RF coils for MRI High-Performance Computing GPU, Grid, Cloud Antenna and EMI diagnostics/ Advanced fast measurements 0 =0 o 0 =45 o Automated design of standard rad. patt. and sparse arrays phi Automated design of compact and reconfigurable antennas and all that 5

6 OTHER TOPICS Inverse Problems: Tomography and Detection Inverse Problems: Antenna Synthesis Simulation of radiolink in hypersonic (re-entry) vehicles Domain-Decomposition and Discontinuous Galerkin Non-deterministic structures 6

7 INVERSE PROBLEMS Find some interesting properties of matter inside a region of interest by -probing the region with electromagnetic fields -or analyzing endogenous fields Examples: medical imaging (ideally, like X-rays, but non-ionizing) medical diagnostics (is there a tumor, a blot clot, brain infarction, is the bone normal or with ostheoporosis ) EEG (endogenous fields) early alert of epileptic seizure (endogenous fields) non-destructive evaluation ground-probing radar (find pipes, de-mining, archeology,.) geological prospection Trivia (but not so much): what you want to see or otherwise sense must have EM properties distinct from the rest 7

8 INVERSE PROBLEMS 8

9 INVERSE PROBLEMS Full mathematical problem, most general ( tomography ) E j H H j E E E, H ( r), ( r), J src r Find the value of permittivity, conductivity (sometime magnetic permeability) from information encoded in the EM fields measured outside the region of interest Non-linear (desperately so ) Ill-posed Usually, trying to get more info than you can gather Still a grand challenge for math, computation and hardware 9

10 INVERSE PROBLEMS Full mathematical problem, most general ( tomography ) Can be viewed as a optimization problem E j H H j E E E, H ( r), ( r), J src r Non-convex, and at each cycle need to solve a full EM problem! 10

11 Approximations INVERSE PROBLEMS E j H H j E E E, H ( r), ( r), J src r Weak scattering: typical in medical, some prospection Born-type approximations linearization, or reduction of non lin Strong difference from rest (e.g. buried metal objects): linear sampling Negligible multiple scatterings: radar, focusing, time-reversal (fast, nearly real time) 11

12 ANTENNA SYNTHESIS (DETERMINISTIC) Standard (equispaced) and sparse arrays; Dynamic-range constrained, upper and lower arbitrary masks Non-linear/ Non convex Still an inverse problem 12

13 SIMULATION OF EM COMMUNICATIONS WITH RE- ENTRY VEHICLES DURING THE HYPERSONIC SHOCK PLASMA PHASE Courtesy of 4.4m = 40 patch antenna frequency = 2.6 GHz (S band) IXV Mach 25 log electron density (plasma freq.) 13

14 EM MODEL/EQUATIONS Plasma medium modeling: Challenges Plasma region is extremely large - Turbulent region (wake) - Wave propagation effects sensitive to gradients of permittivity (tiny refraction index differences over long hauls make rays curve in atmosphere) 10s of meters at some GHz: 100x100x100 wavelengths Numerical solution of full Maxwell eqs. Plagued by instabilities because strong variability and permittivity positive and negative (negative only is ok) (recent findings) Focus: hybrid self-consistent approximate ray solution 14

15 INTEGRAL EQUATIONS WITH DOMAIN-DECOMPOSITION AND DISCONTINUOUS GALERKIN (NON-CONFORMAL MESH) Domain Decomposition: break down a (too) large problem into smaller problems, but solve self-consistently for whole structure Discontinuous Galerkin: different parts of the body have different meshes which are non-conformal ( wrong ) 15

16 NON-DETERMINISTIC PROBLEMS In many problems -the exact geometry is not fully known: fabrication tolerances; natural variability (e.g. biomed) -Or, one wants to try several (silghtly) different geometries Can we obtain the statistics of the desired output without havong to simulate a large number of different realizations? E.g. can we write and equation directly for the average? 16

17 SUMMARY- OTHER TOPICS Inverse Problems: Tomography and Detection Inverse Problems: Antenna Synthesis Simulation of radiolink in hypersonic (re-entry) vehicles Domain-Decomposition and Discontinuous Galerkin Non-deterministic structures 17

18 QUESTIONS? CONTACTS Prof. Giuseppe Vecchi, PhD Antenna and EMC Lab (LACE) Politecnico di Torino Corso Duca degli Abruzzi, Torino (TO), Italy e: w: