HI Murat, > 1) What to do to fix the latency so that it does not change each time > the flowgraph is run? Use a larger-than-worst case fixed delay between the start of RX and TX. GNU Radio is not a "fixed latency" architecture; it's quite the opposite, in fact. > 2) Can I decrease latency below the measured values? Probably, yes. However, latency in a buffer-driven architecture like GNU Radio depends on the buffer sizes, number of connections, and inidividual block latency. So the more complex your flow graph, the higher the latency. > 3) Can B200-mini send ack packets for send calls, when there is no > sob/eob flags are used? I don't understand; could you elaborate on this? If a send call fails, you should get notified. If you don't get a notification of an Underflow or sequence error from the USRP, things are working fine.
Best regards, Marcus On 15.04.2016 09:32, murat tas wrote: > Hi, > > I have a use case similar to the one mentioned in this series of posts: > http://lists.gnu.org/archive/html/discuss-gnuradio/2013-04/msg00421.html > > I am using B200-mini with gnuradio flowgraph containing tx and rx > parts. Transmit part contains a udp source block connected to a > physical layer block(that processes channel encoded bits) and that is > connected to a uhd sink block. Receive part has similar structure with > uhd source, phy block and Socket Pdu block. > > The transmit part should send data continuously to B200-mini. So there > should be signal broadcasted all the time.(Think of it like a base > station broadcast signal) To achieve this, the upper layer of gnuradio > flowgraph sends packets periodically to gnuradio. > > The receive part has very low latency when i measure with a known > signal( from a test instrument). > > When using both transmit and receive parts, I measure the time passed > between the moment a packet enters the udp source block on transmit > side and the moment the same packet sent out from the receive part of > the flowgraph to upper layer. The time passed changes between 150 to > 300 ms for each run of the flowgraph. These numbers are for > GR_FIXED_BUFFER_SIZE decreased to 2048. With default value of 32k, the > time passed is around 2-3 seconds. > > I have three questions: > > 1) What to do to fix the latency so that it does not change each time > the flowgraph is run? > 2) Can I decrease latency below the measured values? > 3) Can B200-mini send ack packets for send calls, when there is no > sob/eob flags are used? > > If anyone previously dealt with a similar use case can offer > alternative methods, it would be great! > > Thanks in advance, > > Murat > > > _______________________________________________ > Discuss-gnuradio mailing list > Discuss-gnuradio@gnu.org > https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
_______________________________________________ Discuss-gnuradio mailing list Discuss-gnuradio@gnu.org https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
