Categorising the sub-mJy population: Star-forming galaxies from deep radio surveys

Models predict that starforming galaxies make up the majority of the source population detected in the very deepest radio surveys. Radio selected samples of starforming galaxies are therefore a potentially excellent method to chart e.g. the cosmic history of star-formation. However, a significant minority of the faintest radio sources are AGN powered 'contaminants', and must be removed from any solely star-formation powered sample. Here we descibe a multi-pronged method for spearating star-forming and AGN powered sources in a deep 1.4 GHz radio survey. We utilise a wealth of multi-wavelength information, including radio spectral and morphological information and radio to mid-IR SED modelling, to select a clean sample of star-formation powered sources. We then derive the 1.4 GHz source counts separately for AGN and SFGs, calculate an independent measure of the evolving star-formation rate density to z∼2, and compare our results to the star-formation rate density determined at other wavelengths.