Hello Jeff Long, Thank you for your answer. I am really new to GNU Radio and DSP, so I am not sure to have understood what you said: do you suggest to get rid of the "stram to vector" blocks ? I also do not know what a single pole IIR filter is, I will document myself and try to implement that solution once I got my head around the concept.
Anyway, thank you for your answer, have a nice day, Alex P. S.: could you give me possible pointers to good references to learn the concepts quicker ? I already stumbled across the wonderful PySDR<https://pysdr.org/> blog, and was wondering if there were other similar/complementary online ressources ? ________________________________ De : Discuss-gnuradio <discuss-gnuradio-bounces+alex-heu=hotmail....@gnu.org> de la part de Jeff Long <willco...@gmail.com> Envoyé : mardi 15 mars 2022 15:43 À : discuss-gnuradio@gnu.org <discuss-gnuradio@gnu.org> Objet : Re: Realise a bandwidth segmentizer The Complex To Mag^2 block can operate on a vector, so that part you can get for free. If you don't actually need the mean, it could be sufficient to use a Single Pole IIR filter (also takes vectors) as a smoothing function. On Tue, Mar 15, 2022 at 9:39 AM e heuchamps <alex-...@hotmail.com<mailto:alex-...@hotmail.com>> wrote: Hello everyone, I am trying to implement a bandwidth segmentizer (cutting a given bandwidth into smaller channels) and energy detector using GNU Radio, as schematically shown here below, but I am experiencing some trouble.[cid:17f8e044bb1cb971f161] More precisely, to achieve the desired result, I have tried to implement two python blocks, given in the attached archive, in files "compute_average_power.py" and "custom_thres.py". The goal of those blocks is to compute the average power contained in all the channels (compute_average_power.py), and feed that computed value to a self-made threshold block (custom_thres.py). So far, the implementation does not work. I have already learned in this post<https://dsp.stackexchange.com/questions/81858/problem-to-extract-value-from-complex-to-mag-squared-gnuradio-block> thanks to Markus Müller that using np.mean (where np is short for numpy) is bad because it depends on the length of input_items. I have however not been able to reach the desired result. Can someone help me out ? Thank you, have a nice day, Alex P. S. : the whole, complete, code is given in the attached archive, in the "bw_segmentizer.py" file.