[en] The dark energy is one of the great mysteries of modern cosmology: it is an unknown component supposed to fill the whole universe and be responsible for the current acceleration of the expansion of our Universe. Its study is a major focus of my thesis: the way I choose to study and characterize this Dark Energy is based on the large-scale structure of the Universe through a probe called the integrated Sachs-Wolfe effect (iSW). This effect is theoretically detectable in the cosmic microwave background (CMB): this light, which originated in the early Universe (380,000 years after the Big Bang), travelled through large structures in the Universe (galaxies and clusters) before reaching us, all of them underlain by gravitational potentials. The acceleration of the expansion (and dark energy) has the effect of stretching and 'flattening' these potentials during the crossing of photons, which has the effect of providing some extra energy of these FDC photons, which will depend on the properties of the dark energy. The iSW effect has a direct but weak effect on the power spectrum of the temperature fluctuations of the FDC effect: it therefore requires the use of external data to be detectable. A conventional approach to this problem is to correlate the FDC with a tracer of the distribution of matter in the Universe (usually galaxy surveys), and therefore the underlying gravitational potential. This has been attempted numerous times with surveys covering a large range of wavelengths, the measured correlation has yet to give a definitive and unambiguous result on the detection of the iSW effect. This is mainly due to the shortcomings of current surveys that are not deep enough and/or have a too low sky coverage. A part of my thesis is devoted to the correlation of FDC with another diffuse background, namely the cosmological infrared background (CIB), which is composed of the integrated emission of the non-resolved distant galaxies. I was able to show that it is an excellent tracer of the gravitational potentials, being free from many of the shortcomings of current surveys. The results of this study shows that the levels of significance for the expected correlation CIB-CMB exceed those of current surveys, and compete with those predicted for the future generation of very large surveys to come (Pan-STARRS, LSST, Euclid). In the following, my thesis was then focused on the individual imprint of the largest structures in the Universe in the CMB by iSW effect. According to an article by Granett et al. (2008), the iSW effect was directly detected by a stacking approach of patches of the CMB at the positions of superstructures. However, the high measured amplitude of the effect seems to be at odds with predictions from the standard model of cosmology. My work on the subject was first involved revisiting this study with my own protocol, completed and associated with a variety of statistical tests to check the significance of these results. This point proved to be particularly difficult to assess and subject to possible selection bias. I extended the use of this detection method to other available catalogues of structures, more consequent and supposedly more sophisticated in their detection algorithms. The results from one of these new catalogues (Sutter et al., 2012) suggests the presence of a signal at scales and amplitude more consistent with the theory, but at more moderate levels of significance than the catalogue and Granett al. At the same time, being a member of the HFI Core Team of the Planck Collaboration, I also performed the same detection using data from the new satellite. The results of the stacking approach introduced a number of questions concerning the nature of the expected signal: this led me to actively work on a theoretical prediction of the iSW effect produced by the superstructures previously mentioned, through simulations based on general relativity and the Lematre-Tolman-Bondi metric. This allowed me to reproduce the structures of Granett et al. and predict the exact full theoretical iSW effect of these structures. This work showed that the central amplitude of the measured signal is consistent with the LCDM theory, but the measured signal shows non-reproducible features that are not compatible with my predictions. An extension of my framework to the additional catalogues that I considered will verify the significance of their associated signals and their compatibility with the theory. Another part of my thesis focuses on a distant time in the history of the Universe, called reionization: the transition from a neutral universe to a fully ionised one under the action of the first stars and other ionising sources. This period has a significant influence on the CMB and its statistical properties, in particular the power spectrum of its polarisation fluctuations. In my case, I focused on the use of temperature measurements of the intergalactic medium during the reionization in order to investigate the possible contribution of the disintegration and annihilation of the hypothetical dark matter. Starting from a theoretical work based on several models of dark matter, I computed and compared predictions to actual measures of the IGM temperature, which allowed me to extract new and interesting constraints on the critical parameters of the dark matter and crucial features of the reionization itself. (author)