Re: [Discuss-gnuradio] GR- IEEE 802.15.4 PHY Issues

[Bastian Bloessl]

The staggering of the imaginary part in combination with the half sine 
pulse shape, makes the constellation point move on the unit circle. 
During a bit period it makes a 90 degree transition to one of the 
neighboring constellations (encoding the 0 and 1). It doesn't actually 
stay at the ideal constellation points, but immediately starts the next 
transition.

Now if the sample rate would be very high, you would see many closely 
spaced points on the unit circle. But the example flow graph uses a 
sample rate of only 4M, i.e., 2 symbols per bit/chip. Therefore, you see 
only one more sample/constellation point when the transition is half way 
through, leading to 8 different points shown in the constellation sink.

Best,
Bastian


On 07/08/2017 09:22 PM, Tellrell White wrote:
> Thanks for the response Bastian
>
> I have taken a look at the IEEE 802.15.4 as suggested. I do see that 
> half sine pulse shaping is used according to the standard. However,  I 
> believe the issue may be with my understanding of O-QPSK. I've attached 
> a picture that I use as a reference. I thought the key difference 
> between qpsk and o-qpsk is that due to the staggering of the imag 
> component by a half a symbol, o-qpsk can only result in 90 degree 
> transitions instead of the 180 degree variations seen in qpsk. I'm still 
> trying to understand why i'm seeing 8 constellation points from my 
> constellation sink and not the usual 4 that you would see in any o-qpsk 
> constellation plot. If any one has a good reference on this please share.
>
> Tellrell
>
>
> On Saturday, July 8, 2017 5:53 AM, Bastian Bloessl <address@hidden> wrote:
>
>
> Hi,
>
> On 07/08/2017 05:03 AM, Tellrell White wrote:
>  > Hello Guys
>  > I'm currently using the IEEE 802.15.4 PHY. I've added a random source to
>  > the flow graph and also a qt gui time and constellation sink. I have two
>  > questions, why does the constellation plot attached show more than the
>  > usual 4 constellation points seen in O-QPSK constellation?? Is it due to
>  > the sampling rate I'm using which is 8M in this case.
>
> The default sampling rate is 4M, I don't know what you changed when you
> tried to make it 8M. If you add a constellation sink to the unmodified
> flowgraph you will see constellation points on the unit circle (your
> second screenshot), which is what you would expect, since IEEE 802.15.4
> uses O-QPSK with a half-sine pulse shape (see the standard).
>
>
>  >
>  > Also, I was curious as to if the vector source added which implements
>  > the half sine pulse shaping was creating a problem so I disabled it and
>  > I was able to see the expected 4 constellation points. So, is the source
>  > of extra constellation points I'm seeing?? I've attached both the .grc
>  > file and 2 screen shots, one representing both constellation plots
>  > discussed.
>
>
> With this you removed the pulse shaping, so you see only the QPSK
> constellations, but it's also not 15.4 anymore. Most standards use some
> kind of pulse shaping, so you don't see only the constellation points
> when you look at the final IQ signal. I don't understand why this is a
> problem for you.
>
>
> Best,
> Bastian
>
>
>
>
>
> _______________________________________________
> Discuss-gnuradio mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

--
Dipl.-Inform. Bastian Bloessl
CONNECT Center
Trinity College Dublin

GitHub/Twitter: @bastibl
https://www.bastibl.net/