The importance of making accurate measurements of the fluctuations in the CMBR is now widely appreciated. Indeed, by making maps of these fluctuations and by measuring their power spectrum, it is hoped that tight constraints may be placed on fundamental cosmological parameters and that we may distinguish between competing theories of structure formation in the early Universe such as inflation and topological defects.
Several ground-based and balloon-borne experiments are planned over the next few years, and these should provide accurate images of the CMBR fluctuations and lead to a significant improvement in the measurement of the CMBR power spectrum. Nevertheless, these experiments are unlikely to be able to achieve the accuracy required to resolve numerous degeneracies that exist in the parameter set of, for example, the standard inflationary CDM model. As a result, a new generation of CMBR satellites are currently in the final stages of design, and it is hoped that these experiments will provide definitive measurements of the CMBR power spectrum as well as detailed all-sky maps of the fluctuations.
According to current estimates, the NASA MAP satellite is due to be launched in 2000, followed by the ESA Planck Surveyor mission in 2005. Both experiments aim to make high-resolution, low-noise maps of the whole sky at several observing frequencies. As with any CMBR experiment, however, the maps produced will contain contributions from various foreground components. The main foreground components are expected to be Galactic dust, free--free and synchrotron emission as well as the kinetic and thermal SZ effects from galaxy clusters. In addition, significant contamination from extragalactic points sources is also likely.
Aside from extragalactic point sources, the other physical components affecting any CMB experiment have reasonably well defined spectral characteristics, and we may use this information, together with multifrequency observations, to distinguish between the various foregrounds. Several linear methods have been suggested to perform this separation, many of which are based on Wiener filtering (e.g. Bouchet et al. (1996); Tegmark & Efstathiou (1996); Bouchet et al. (1997)). In this paper, however, we investigate the use of a non-linear maximum entropy method (MEM) for separating out the emission due to different physical components and compare its performance with the Wiener filter approach. We apply these methods to simulated observations from the Planck surveyor satellite but, of course, the same algorithms can be used to analyse data from the MAP satellite. The application of the MEM technique to simulated interferometer observations of the CMBR is discussed in Maisinger, Hobson & Lasenby (1997) and the method has also been applied to the analysis of ground-based switched-beam observations by Jones et al. (1998).