Hi everyone,
I am interested in knowing the delay jitter of the total transmission
time of a packet/waveform. Specifically, I want to know the time
between the time the flowgraph is tsarted in python (tb.run or
tb.start) and the time that the first sample is transmitted into the
air from the USRP hardware. I want to know if I can reduce the jitter
in this time (across runs) to as low a value as possible.
I use usrp_siggen.py with gnuradio-3.1.2 to transmit a square waveform
.I try to start the flowgraph at a precise time x (in microseconds) by
using y=x-time.time() and time.sleep(y) and then
tb.start(),time.sleep(0.1) and tb.stop(). I also use an interp of 32 at
the Tx. I have a receiver that logs all data (with -d 64). I observe
the samples just out of the usrp source.
For transmissions that are precisely spaced in time using appropriate
values x, the inter transmission delay measured at the receiver
is off by a value > than 2 ms.
(despite using nice to increase the priority of this process).
1. Is it possible to reduce the jitter between successive Tx.delay measurements to be few 10's of microseconds or lesser?
2. Is there a way to run the usrp_siggen code as a kernel module to improve the delay jitter performance?