Investigation and report on Polynomial Baseline Dependence Averaging

PBDA (Polynomial Baseline Dependent Averaging) could become a useful technique to:

1. Average visibilities at the correlator end, thus reducing bandwidth requirements. Early studies show that a reduction of between 2 and 4-5 is possible according to the case.
2. Average visibilities before the gridder. A greater compression rate could be achieved.  Preliminary studies show between 3 and > 10 according to the case.

Some work mostly on standard BDA had already been carried out under SDP and published by S.Salvini, S.Wijnholds et al.

Here we intend to produce a new document and preliminary prototype software, based on the results already obtained, to show the usefulness both in theory and practice to employ PBDA in SKA to reduce costs considerably.