Abstract :
[en] With the growing number of wireless devices and connected objects, the level of electromagnetic noise could reach unprecedented levels, potentially leading to malfunctions. In order to limit this effect, it is important on the one hand to develop systems with enhanced immunity to radiated electromagnetic fields and, on the other hand, to further rationalize the electromagnetic radiation of wireless devices by improving their design.
In this dissertation, two axes of research are addressed. The first aims to help prototyping antennas by accelerating and simplifying the measurement of antenna radiation patterns by performing this measurement in non-dedicated environments, i.e. outside fully anechoic chambers. To achieve this, a method of correction is applied to measured electromagnetic fields a posteriori to remove from the radiation pattern the spurious contributions introduced by reflections occurring in the reverberating test environment. The method described in the first part of the manuscript is based on the angular deconvolution of the impulse response of the environment from the echoic pattern. At the end, the results of experimental validations, obtained through several test cases, are presented.
The second axis focuses on the electromagnetic field shaping in reverberating and complex environments, as it would be particularly useful for radiated immunity electromagnetic compatibility tests, allowing them to be performed in arbitrarily reverberating environments without requiring the systematic use of anechoic chambers. Field shaping techniques aim on the one hand to suppress any electromagnetic field introduced by one or more sources of undesired noise in a determined part of an arbitrary environment, and on the other hand to superimpose any arbitrary electromagnetic field map such as plane waves used for radiated immunity tests. To achieve this, a method of field shaping, based on the optimal control theory of Maxwell's equations, is presented. The potential of this method can be appreciated thanks to several numerical simulations presented for various environments. Then, by means of two experimental test cases of more modest complexity, the experimental validité of the method is established.
