Sonohi / monster

MONSTeR is a framework built around the LTE system toolbox available in Matlab
MIT License
23 stars 9 forks source link

Wrong scaling of Es/N0 for frequency type interference #126

Open jakthra opened 4 years ago

jakthra commented 4 years ago

Computing the interfering waveform from interfering sources uses a normalization power scaling. This power scaling is required to "normalize" the values for the receiver chain such the energy per symbol is kept at a constant magnitude of 1. Currently the interference is computed as follows:

  1. Sum of the interfering waveforms with the respective power levels (signal power levels)
  2. The interfering waveform is then added to the scaled waveform (scaled by the power level as dictated by the link budget)
  3. The final combined waveform (with added interference) is then scaled back to the original power level

In other words, the interference is added per signal power and not per symbol energy. This means that the symbol energy is changed at the final received waveform. This is due to the energy being distributed onto the final waveform. This results in final symbol energy that is lower than 1.

The fix should compute the symbol energy (and the N0) for proper scaling magnitudes.