I study GMSK modulator. In its simulation we use an oversampling ratio if we implement gaussian filter or oversample the input signal.

In my simulation I can test different value of the oversampling ratio;2, 4, 6 and so on... but honestly, I don’t see a different and no idea how to determine the best ratio for my simulation.

could you explain how you determine the oversampling rate? is there an equation to do that?

PS it is not given sampling/Nyquist frequencies, so it is impossible for me to compute its ratio to find the oversampling ratio

  • 1
    $\begingroup$ well if you have no external constraints on sampling frequency, you can choose whatever works. You have identified exactly the thing that constrains your oversampling ratio, and excluded it, so there's nothing left. $\endgroup$ May 16, 2022 at 9:44
  • $\begingroup$ @MarcusMüller What characteristics can limit my choice? $\endgroup$
    – FrimHart64
    May 16, 2022 at 10:48
  • 1
    $\begingroup$ as said, mostly sampling rate, and as Dan explains, that's a system level design decision. $\endgroup$ May 16, 2022 at 10:50

1 Answer 1


The choice of oversampling ratio is a system level decision where "system" includes the complete physical layer modulator from symbol mapping to the analog output. There will be a minimum oversampling ratio sufficient for waveform quality, and for GMSK with BT=0.5 or higher, 4 samples per symbol will be more than sufficient, as detail in this link. Beyond that, higher oversampling ratios will require more processing but will simplify subsequent filtering, ultimately the analog reconstruction filtering after the D/A conversion. Such a decision is made with these factors in mind and for the transmitter the out-of-band emissions is often the driving requirement.

Further oversampling can also be needed for power amplifier predistortion techniques when used, since a much wider bandwidth is needed to measure and compensate for the non-linearity that the power amplifier creates. This would not be a consideration for GMSK given the constant envelope modulation can be used with highly non-linear power amplifiers.

  • $\begingroup$ in the post you linked, you explained, the oversampling ratio =4 is sufficient for gmsk modulation. For a transmission the modulated signal( gmsk output) I have a transmitter with max datarate 60mbps. for this data rate , 4 samples is wrong choice, and I can not understand why it is. Oversampling ratio descreases the data rate, in 4 times...but how it can be a reason i can not use 4 samples? $\endgroup$
    – FrimHart64
    Jun 8, 2022 at 6:20
  • $\begingroup$ @FrimHart64 This depends on the BT used. The referenced link was with BT = 0.5, and for that 4 samples per symbol is sufficient as shown in the link. I'll update this answer to make that clearer. $\endgroup$ Jun 8, 2022 at 11:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.