# How to design a filter with smoothness regularisation?

I am rather new to the DSP topic. However, I would like to filter a signal with lowest error and highest smoothness.

Until now, I was using tikhonov regularisation inside batch regression with some kind of kernels.

y=Ax, with x= (X[0], ... X[t])^T


The issue is however I do not know how to assess the error in the frequency domain then.

Could anyone advice where I might look for a discussion on this subject?