Re: [Discuss-gnuradio] ber_simulation

[Marcus Müller]

Hi Ekko,

i got ber is not zero

as discussed several times now, you will never get BER in any system with a real radio.

So, there's a lot going on here:
* delay, and equivalently
* phase shift (how do you know your PSK system doesn't decode with a wrong, random phase?),
* attenuation,
* quantization,
* unwanted intermodulation in (those should be small, normally, but in a direct cable system, it's very easy to drive RX into saturation, making these important)
  * TX amplifiers
  * TX mixer
  * RX amplifier
  * RX mixer
* possibly even frequency-dependent behavior (not very likely in a cable),
* some additive noise

So, in real world receivers, there's a lot more than the demodulation using the constellation decoder -- there's equalization, timing recovery, usually also frequency recovery, proper "start of signal" detection, channel coding, and much more to make a receiver deal with what the real world does to the transmitted signal.

You're not trying any of that. Most importantly: you're really not understanding what Derek was referring to when he said there's delay:
Your receiver side simply can't know how much delay is there. You will have to build something that estimates the phase shift that happened. Otherwise, a PSK will decode something that is a rotated version of the constellation sent.

This is all very basic radio theory. It would really be good if you could find the time to sit back with a text book on digital communications. What you're doing right now is really digging in the dark; with the proper theoretical background you'd be able to recognize and solve the problems you're encountering yourself. If you continue you to work like this, it will take you very long to always come here and ask people. In the end, you will have more success, fun, and speed, if you take the time now to improve your theoretical knowledge.

Best regards,
Marcus

On 18.11.2015 13:21, chai E wrote:

hello derek

because of the delay of actual radio,i change the data sourece into a static number 1,and use bpsk

​
and set the uhd source gain into 10,i got ber is not zero ,i check the source ,the number in two file is different.

but i change to transmit 3 with qpsk,the ber is not zero

​

so i think it is not the matter of actual radio delay.

you said add some packeting,is it mean the block "repackbits"

and what the flow control of sync mean? which block

thank you

--Ekko
​

2015-11-17 23:18 GMT+08:00 Derek Kozel <address@hidden>:

Hello Ekko,

In the first simulation there is no delay or channel effects, other than a small amount of simulated noise, so the data bits line up perfectly at the start of the error rate block. Now that you are transmitting over actual radios you have to account for all of the additional delay. That is your correct bit error rate right now, you are just not measuring what you expect. You will need to, at least, add packetising and some flow control to synchronize the inputs to the error rate block.

Regards,

Derek

On Tue, Nov 17, 2015 at 3:04 PM, chai E <address@hidden> wrote:

today i run the /usr/share/gnuradio/examples/digital/demo/ber-simulation.grc

just like 

the picture1, i exe the grc file and got the ber is near zero,i think this is normal

then i seprate this grc into wo part 

the one is chunks to symbols+uhd_sink

the other is uhd_source+constellation decoder,

disable the noise,

the result is 

it is too large ,this is symbol error rate。

next is my set of uhd

​

center_freq is 433M

i want to know is there something wrong in my the uhd setting,

and why the symbol error rate is so large.

​

hope i have Elaborated my problem  clearly

--Ekko
 ​

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio