Abstract : This thesis is motivated by upcoming high-resolution, high-sensitivity microwave background experiments, which should be sensitive to the CMB polarization and lensing. The first chapter provides a review of the CMB polarization with emphasis on future related experiments. The Planck experiment is described in a second chapter, where I develop a fast simulation code of Planck time-ordered data optimised to ease elaboration and test of data analysis methods. The two last chapters deal with gravitational lensing of the cosmic background radiation. First, I evaluate the capability of the upcoming experiments mentioned above to measure the power spectrum of Large Scale Structure by means of the extraction of weak lensing. Then I derive their sensitivity to the total neutrino mass, using the suppression of power due to free-streaming of massive neutrinos. Finally, I develop a method to estimate the foregrounds effects in the gravitational lensing extraction process. This method uses the best linear estimator avalaible in the litterature and is validated by numerical simulations that include non-gaussian CMB lensed maps and extra-galactic radio sources maps. I find that sources emission reduces the sensitivity of future experiments to the weak lensing and leads to an overestimate of the convergence power spectrum.