The paper is devoted to the methods of determination of the cosmological parameters from recent CMB observations. We show that the more complex models of kinetics of recombination with a few "missing" parameters describing the recombination process provide better agreement between measured and expected characteristics of the CMB anisotropy. In particular, we consider the external sources of the Ly-{alpha} and Ly-{c} radiation and the model with the strong clustering of baryonic component. These factors can constrain the estimates of the cosmological parameters usually discussed. We demonstrate also that the measurements of polarization can improve these estimates and, for the precision expected for the PLANCK mission, allow to discriminate a wide class of models.

We develop a linear algorithm for extracting extragalactic point sources for the Compact Source Catalogue of the upcoming PLANCK mission. This algorithm is based on a simple top-hat filter in the harmonic domain with an adaptive filtering range which does not require a priori knowledge of the CMB power spectrum and the experiment parameters such as the beam size and shape nor pixel noise level.

We present a new method to extract the beam shape incorporated in the pixelized map of CMB experiments. This method is based on the interplay of the amplitudes and phases of the signal and instrumental noise. By adding controlled white noise onto the map, the phases are perturbed in such a way that the beam shape manifests itself through the mean-squared value of the difference between original and perturbed phases. This method is useful in extracting preliminary antenna beam shape without time-consuming spherical harmonic computations.

We discuss some aspects of the Cosmological Gene Project started at the Special Astrophysical Observatory (Russia) in 1999. The goal of the project is to measure the anisotropy and polarization of the Cosmic Microwave Background (CMB) and investigation of atmospheric fluctuations and foreground on arcmin-scales using the radio-telescope RATAN-600. We develop the cluster analysis of one-dimensional random fields for the application to the RATAN-600 scans. We analyze the specific properties of peak clusterisation in the RATAN-600 scans which to separate the primordial CMB signal from noise.