On Mon, Jun 13, 2011 at 17:06, Marcus D. Leech <[email protected]> wrote:
> As long as overruns aren't happening *frequently* (like more than once > every few dozen higher-layer-data-frames), then one can regard > such interrupts in data flow as equivalent to channel noise, which is > something you have to cope with *regardless*--even if your > sample stream is "perfect". > > In fact, if your channel model makes brash assumptions like the only > "noise" will be from the "natural physics" of the channel, then you've > already entered a state of sin. In general, radio channels aren't > well-behaved with regard to "goop" on the channel, and I regard sampling > overruns, at a 10,000ft level, as broadly-equivalent to "channel goop". > To a receiver, these appear as instantaneous phase jumps--which can, for some modulations, take a long time to recover from. At least now with UHD one can update the receiver's notion of sample number/timebase accurately when such a gap occurs. Johnathan
_______________________________________________ Discuss-gnuradio mailing list [email protected] https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
