Thanks for the responses.
In the decimation case we're considering an already sampled signal and wish to reduce the frequency by only considering every Mth sample, thus decimating by M.
In undersampling I guess you rather have a continous signal and sample it a frequency lower than the signals highest frequencies (but still >2xBW). So an IF signal centered at 45MHz and sampled at 20MHz will alias to 5MHz.
So in both cases we're reducing the frequency of the signals. Should be pretty much the same. Am I wrong here?
Cheers,