Why not fractional space spampling when frequency domain equalization?

as we all konw, fractional space(FS) spampling is robust to nonideal timing
synchronization when we do time domain decision-feedback equalization. My
question is: why no paper mentioned using FS frequency domain DFE
equalization? at least till now I havenot read a paper confirm it. Any
comment or reference will be helpful to me. Thanks!