At present, gr-dpdpostdistorter produces an estimation for every corresponding value of PA_input and PA_output which it recieves but this can lead to over-fit or under-fit of the coefficients. For avoiding this situation we need to incorporate methods in the current implementation or modify it in a way such that each estimation is performed for atleast several hundred samples of corresponding PA_input and PA_output data to achieve best fitting coefficients estimated for them.
At present, gr-dpd postdistorter produces an estimation for every corresponding value of PA_input and PA_output which it recieves but this can lead to over-fit or under-fit of the coefficients. For avoiding this situation we need to incorporate methods in the current implementation or modify it in a way such that each estimation is performed for atleast several hundred samples of corresponding PA_input and PA_output data to achieve best fitting coefficients estimated for them.