Hello everbody,

upsampling a complex data stream gives me headaches :( In my flowgraph the upsampling is done by multiplexing 7 zeros with 1 data sample and feeding this into a FIR filter, which is how everybody does it as I understood. However, for further processing I have to take the delay into account, that is introduced by the filter (length N). This delay should be (N-1)/2, as the filter phase is linear. I do this by delaying the data stream by one symbol length minus (N-1)/2, since causality rules.

       +----+  ------------+
       |Data+--->          |
       +----+   |          |   +---------------------+   +-----+
                |Stream MUX+--->Decimating FIR Filter+--->Delay|
+-----------+   |          |   +---------------------+   +-----+
|Null Source+--->          |
+-----------+  ------------+

If I upsample as depicted above, my transmission kinda works, but I end up having alignment problems resulting in a lower SNR. As I understood the delay is realized by zero padding, which should work, but there seems to be something I am missing out on. When I use the Decimating FIR Filter block and I want the same alignment as before (symbol x starts at sample y), so it can be further processed, what do I have to do?

Thank you for your ideas.

Regards,

Max

_______________________________________________
Discuss-gnuradio mailing list
Discuss-gnuradio@gnu.org
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

Reply via email to