Recent progress in sub-millimeter wave receiver technology gives the possibility to drastically improve the quality of limb sounding data by use of a superconductor-insulator-superconductor (SIS) mixer element. This receiver will detect molecular spectra with a signal-to-noise ratio one order of magnitude better than the conventional ambient-temperature Schottky receiver. SMILES (Superconductor Submillimeter-wave Limb-emission Sounder) is proposed by the Communications Research Laboratory and the National Space Agency of Japan, with technical support from the National Astronomical Observatory, and with scientific support from the University of Bremen, in order to demonstrate the new sub-millimeter wave technology in space, and to conduct the measurements of limb-emission sounding for a group of molecular species profiles. In order to anticipate the performance of the instrument, retrieval simulations are carried out. Synthetic measurements, as will be recorded by the SMILES instrument, are generated by the use of a forward model. These are then inverted, using an inversion model, in order to derive the variables of interest, such as molecular species profiles (e.g., O3, ClO, HCl), atmospheric temperature profile, or a first order instrumental pointing correction (i.e., a pointing offset). The applied inversion algorithm is the optimal estimation method (OEM). The advantage of the OEM is that it allows a formal error analysis needed for a general error characterization of retrieval performance. The error analysis takes into consideration the total statistical error, the measurement error, the vertical altitude resolution, and the correlation between the retrieved quantities. The altitude domain of a good measurement sensibility is defined by the measurement response.