u sample the data with f < fmax , this will result in the translation of the signal to baseband , but to prevent aliasing u must keep f > 2B (where B is the channel BW) , but this technique suffer from the noise folding
In a simple words, isn't a sub-harmonic mixer a mixer, where LO signal's frequency is a sub-harmonic of the LO signal's frequency of the conventional mixer?
Suppose you mix RF 10GHz with LO 9GHz (IF 1GHz) with conventional mixer. For sub-harmonic mixer the LO would be, let say, 3GHz and you would still get IF of 1GHz due to the hamrnonic mixing.