Hm, that is an interesting result. The point is that the polyphase "magic" that allows decimation before pushing samples through a FIR is mathematically 100% equivalent to doing the decimation after the FIR. Nearly the same goes for (non-decimating) FIR vs FFT filter: whereas the FIR really just "pushes" the samples through convolution with the taps in time domain, the FFT filter just multiplies the frequency domain representation of blocks of samples with the frequency domain equivalent to the taps, before transforming the result back to time domain. Should be absolutely identical.
Now, the question is: if the filters were functionally actually 100% identical, what would explain the different FER? Or are we comparing the Flowgraph pre-rectructuring with the flowgraph-post-restructuring, where other things have changed, too? Best regards, Marcus On 17.12.2015 12:55, Saulo Queiroz wrote: > Hi > > In my "through-the-air" tests with a couple of B210s the FFT filters > presented much lower > FER in comparison to the FIR filters at the Rx side. According to [1], the > "FFT filters" > downsamples after filtering while FIR downsamples before filtering. > > cheers > > [1] > http://www.trondeau.com/blog/2014/2/27/to-use-or-not-to-use-fft-filters.html > > On 15 December 2015 at 19:06, Saulo Queiroz <[email protected]> wrote: > >> Many thanks Bastian! >> >> I checked the loopback version and it seems to work very well. >> >> I'll check over-the-air and report here! >> >> On 15 December 2015 at 19:04, Bastian Bloessl <[email protected]> >> wrote: >> >>> Hi, >>> >>> I replaced the Frequency Xlating FFT filters with FIR filters, used the >>> Low-Pass Filter Taps block to generate taps (since I can’t get my head >>> around this notation), and removed the filter from the first conversion. >>> >>> Now, it seems to work. At least it receives frames. If you still have >>> problems I can send you the flow graph. >>> >>> Best, >>> Bastian >>> >>> >>>> On 15 Dec 2015, at 10:05, Saulo Queiroz <[email protected]> wrote: >>>> >>>> Each stream has been shiffted with the xlating block. >>>> The intention is to "split" a 20 MHz wide analog channel into two of 10 >>> MHz. >>>> Each 10 MHz channel transmit its own ofdm frame. >>>> I attached the flowgraph for more details. >>>> >>>> thanks in advance. >>>> >>>> On 15 December 2015 at 17:42, Martin Braun-2 [via GnuRadio] <[hidden >>> email]> wrote: >>>> tP indicates you're using corrupt tagged streams, maybe your add block >>>> is overlaying them? I'm also not entirely sure what you mean by >>>> 'simultaneous parallel transmissions'. Are they on different >>>> frequencies? Are you mixing them together in baseband? >>>> >>>> Cheers, >>>> Martin >>>> >>>> >>>> On 15.12.2015 04:10, Saulo Queiroz wrote: >>>> >>>>> Hi all, >>>>> >>>>> I'm trying to Tx a same tagged stream simultaneously through two >>> analog >>>>> orthogonal channels. >>>>> >>>>> The flow path of each stream copy is: resampling, adjust tag lenght >>> and >>>>> xlating FFT filter (with shifting). After this I take the output of >>> each >>>>> filter and put into and add block then to the USRP sink. I also do the >>>>> reverse process at the Rx side. >>>>> With some packets are successfuly receive but with so many losses. At >>>>> the Tx side I get many "tP". Any tip on how to set simultaneous >>> parallel >>>>> transmissions without this? >>>>> >>>>> I'm using gr-ieee80211 (thanks Bastian and team :) that has worked >>>>> nicely with the single channel scenario. >>>>> >>>>> thanks in advance >>>>> >>>>> BR >>>>> >>>>> -- >>>>> Saulo Jorge bq >>>>> - "Beware of bugs in the above code; I have only proved it correct, >>> not >>>>> tried it." >>>>> Donald Knuth. >>>>> >>>>> >>>>> _______________________________________________ >>>>> Discuss-gnuradio mailing list >>>>> [hidden email] >>>>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio >>>>> >>>> >>>> _______________________________________________ >>>> Discuss-gnuradio mailing list >>>> [hidden email] >>>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio >>>> >>>> >>>> If you reply to this email, your message will be added to the >>> discussion below: >>> http://gnuradio.4.n7.nabble.com/Tag-preemption-USRP-sink-tp57286p57297.html >>>> To start a new topic under GnuRadio, email [hidden email] >>>> To unsubscribe from GnuRadio, click here. >>>> NAML >>>> >>>> >>>> >>>> -- >>>> Saulo Jorge bq >>>> - "Beware of bugs in the above code; I have only proved it correct, not >>> tried it." >>>> Donald Knuth. >>>> >>>> wifi_tx_rx_loopback.grc (134K) Download Attachment >>>> >>>> View this message in context: Re: Tag preemption USRP sink >>>> Sent from the GnuRadio mailing list archive at Nabble.com. >>>> _______________________________________________ >>>> Discuss-gnuradio mailing list >>>> [email protected] >>>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio >>> -- >>> Dipl.-Inform. Bastian Bloessl >>> Distributed Embedded Systems Group >>> University of Paderborn, Germany >>> http://www.ccs-labs.org/~bloessl/ >>> >>> >> >> -- >> Saulo Jorge bq >> - "Beware of bugs in the above code; I have only proved it correct, not >> tried it." >> Donald Knuth. >> > > > > > _______________________________________________ > Discuss-gnuradio mailing list > [email protected] > https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
_______________________________________________ Discuss-gnuradio mailing list [email protected] https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
