[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] Power control for QAM demodulate

From: Marcus Müller
Subject: Re: [Discuss-gnuradio] Power control for QAM demodulate
Date: Sat, 18 Feb 2017 17:26:43 +0100
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.2.0

To ease answering, I'll add more of the conversation in this separate mail:

You wrote:

Thanks so much for your advice. I have read the PAPR OFDM document and set the higher RX and TX gain at both transmitter and receiver. But I still cannot demodulate the signal.
Yes, because it's intrinsically harder to demodulate 16QAM compared to QPSK, as I tried to explain.
So my question is do I need to change some parameters or adding some algorithms to control the USRP to intelligently receive different signal with different power.
These signals do *not* have a different power per se. SNR should be the same.
Has anyone achieved such function before?
Or we can not use the USRP to receive the QAM signal.
Of course you can! It's really just a question of choosing transmission parameters so that things work
I actually don't exactly know how USRP works to receive the signal and convert it to the complex number.
You pretty much said it: USRPs (in general) are direct mixing complex baseband receivers. There's a lot of literature on how these work!
I actually also tested the signal carrier system,  but 16QAM still not works.
I'm afraid you'll have to dig into what noise is, how it affects demodulation, and where symbol errors come from. This is theory that you'll need to know when modifying/implementing digital transceiver systems!

Best regards,


On 02/18/2017 05:22 PM, Marcus Müller wrote:

Hi Shangqing,

I think you forgot to mention that you've already gotten answers to this on usrp-users:

I wrote:

Anyway, yes, 16 QAM is of course a lot more prone to noise than QPSK, when using the same average power. (that's kind of logical – the 16 QAM constellation points are a lot closer together than the four QPSK points)

With any modulation, you'd set the RX gain as high as possible without clipping – and that is a real problem with OFDM. I recommend you google "OFDM PAPR". So, you'd use the same RX gain for QPSK as for QAM modulation.

You'd of course also use the highest possible distortion-free TX gain, for both modulations, again, the same.

So, what you see is exactly what you've learned in digital comms 101 – the closer constellation points are, the more likely it is you get a symbol error. What did you expect?
And Kevin wrote:
Also keep in mind that in case of QPSK, the receiver does not need to correct for amplitude. The receiver only needs to correct the phase distortion caused by the channel since all information is encoded in the phase of the QPSK symbols. 

However, in 16-QAM modulation, information is encoded in both amplitude and phase of the 16-QAM symbols. Now receiver needs to correct both amplitude and phase distortion caused by the channel.

So I am not sure if the receiver algorithm can handle QAM.

So, instead of asking the same question here, I'd recommend you address what you did not understand in the answers you've gotten this far! It'll make it much easier for us to help you.

Sadly, you haven't addressed any of the points we made – neither that the average power is, in contrast to what you claim, the same for QAM and QPSK, nor the additional hardness in demodulating these finer-grained constellations.

Best regards,

On 02/18/2017 04:23 PM, Zhao Shangqing wrote:

Hi all,

I am using Gnuradio and USRP X300. I am implementing OFDM communication using 16 or 64 QAM modulation scheme. I am using the example of OFDM given by Gnuradio, and just changing the payload modulation to 16QAM. I cannot correctly demodulate all the samples, but QPSK works fine. I know QAM has the different signal power. I wish to know if I wish to demodulate the M-QAM signal, how to control the receiving power to calibrate different signals to different powers. 

Best regards,


Discuss-gnuradio mailing list

reply via email to

[Prev in Thread] Current Thread [Next in Thread]