While DARWIN needs a ground based demonstrator, this preparatory mission is better carried out by an integrated experiment (optimized at the system level) than by an instrument constrained by an existing infrastructure. Because the quality of the site (seeing, emissivity) is the main contributing factor to an interferometer's performance, this experiment should be located on the best accessible site. We argue therefore that the ground based preparation to DARWIN should be carried out from the top of the ground turbulence layer on the Antarctic plateau. Simulations on a possible antarctic instrumental concept (ALADDIN) show that a pair of even relatively modest collectors (1m) on a small baseline (up to 40m) are sufficient to achieve a sensitivity (in terms of detectable zodi levels) which is about twice better than that of a nulling instrument on a large interferometer (such as GENIE at the VLTI), and to reach the 20-zodi threshold value identified to carry out the DARWIN precursor science.