please help me with QPSK modulation

[suribright]

Hi,

I would like to run the QPSK at 13.56 MHz of carrier frequency. Any one have a matlab code please send me one. I have tried to write this code but did not work at 13.56 MHz. Wont know what wrong is my code?. In my code, I write the carrier at 13.56Mhz and sampling frequency at 120KHz. When I take FFT of signal it always show at 40KHz. I would like to show FFT at 13.56 MHz. Any one have know about QPSK please let me know or if you have a code it is really appreciated. Thanks

 

[ChakerQ]

Hi,

Post the code and i will help you

 

[suribright]

Just message you my code, Mohamed.

 

[ChakerQ]

Hi,

I send you a code by message. Plz give me your feedback.

bye

 

[suribright]

Hi,
I created a m file in matlab but not work by your code. It shows the power spectrum at lest than 10Khz. Not at 13.56Mhz. I saw your explanation before but it closed now. Dont know why. Can you show it again?. If i run my code at sampling frequency 120Mhz, it will show FFT at 13.56Mhz. But I dont want to sample high frequency. I want sampling at 120Khz. waiting for your reply.Tks.

 

[ChakerQ]

Hi,

This is the old reply.. i delete it because i coded it and i send you the code.

i advise you :

1. You must precise fc (carrier frequency), fs (sampling frequency) and B (baud rate). Then, N=fs/B will be the number of sample in one symbol of period T=1/B;

2. Time vector t=0:1/fsN-1)/fs; % time for one symbol

3. M=4; % 4 state for QPSK modulation
k=log2(M)=2; % 2 bits for ones symbol
data=randint(1,k*Nsym,[0 1]); % generate random binary vector [0 1 1 0 1 0 1 1 0 0 .... ]
qpskdata=bi2de(reshape(data,Nsym,k),'left-msb'); % Convert the bits in data into k-bit symbols [1 2 2 3 0 ...], 00=0 / 01=1 / 10=2 / 11=3

4. you must code data in phase. you have choice phi(i)=2*pi*qpskdata(i)/M [0,pi/2,pi,3pi/2] or phi(i)=(2*qpskdata(i)+1)*pi/M [pi/4,3pi/4,5pi/4,7pi/4]
your vector will be like that [pi/2 pi pi 3pi/2 0 ...] : QPSK constellation
finely : s(i)=A*sin(2*pi*fc*t+phi(i))

y= []
for i=1:Nsym
y=[y A*sin(2*pi*fc*t+phi(i))]
end

Concerning your question.
When you sampled your signal at high frequency (Fs=120Mhz > 2*(fc+B/2)), you are performing lowpass signal sampling.
Then, in your spectrum [0-->Fs/2] you can see your signal at fc frequency (fc<Fs/2).

When you sampled your signal at low frequency (Fs=120kHz > 2*B), you are performing bandpass signal sampling.
In this case, you can't see signal at fc frequency because fc is greater than your spectrum fc>Fs/2.
But, you will see a replica of your signal at fc1=rem(fc,fs)=B/2.
Because like we know the spectrum of a sampling signal is a replica of the continuous signal at each Fs interval [0 Fs] [Fs 2FS] [2Fs 3FS] [3Fs 4FS] ...
so fc will be at one interval [k*Fs (k+1)*FS], and this spectrum is the same as [0 Fs/2] represented.

look for this link.. all this is well described (not like my poor English lol)
https://www.informit.com/articles/article.aspx?p=345472&seqNum=3

best regards

 

[suribright]

So please show me that spectrum at one interval [k*Fs (k+1)*FS] not in [0 Fs/2]. So we can see the spectrum exactly at 13.56Mhz. I have tried that but not work. Please write in my code if it's possible. You can assume B=10 Khz. Tks

 

[ChakerQ]

Hi,

I send to you the new code where you can see the spectrum with the lobe at fc.

I will check your code but i can't understand why you are using I & Q branch.

 

[suribright]

Hi,

Oh just because I just want to show how IandQ branch works in process. In your code I think you used bandwidth =fs/N=120e3/1024=117Hz. Is this right? It's small bandwidth I think.

 

[ChakerQ]

Hi,

Yes i used N=1024 to have N points in the spectrum. You can change N to have a large bandwidth up to fs/2=60kHz. Or to consider the N samples for more than one symbol.

I you want to have the same spectrum with your code. You must :

1. verify bandpass sampling

2. simulate with only one symbol

 

[suribright]

Hi,

Dont understand below part in your code?. What mean of k1, kk, fc?.

N_cyc=(N*fc)/fs; % Number of cycles
k1=round(N_cyc);
kk=0;
k1=k1-1;
while(kk==0)
k1=k1+1;
kk=isprime(k1);
end
fc=(k1*fs)/N; % coherent carrier frequency

fc1=rem(fc,fs); % shiftted fc
k_fs=floor(fc/fs);
kfs=k_fs*fs;

Do you know any books about modulation scheme (QPSK)?. I would like to have a look. Tks

 

[ChakerQ]

Hi,

_cyc=(N*fc)/fs; % Number of cycles
k1=round(N_cyc);
kk=0;
k1=k1-1;
while(kk==0)
k1=k1+1;
kk=isprime(k1);
end 
fc=(k1*fs)/N; % coherent carrier frequency

This part of code used to change frequencies fs and fc to coherent forms. It will gives a best spectrum.

https://en.wikipedia.org/wiki/Coherent_sampling

fc1=rem(fc,fs); % shiftted fc 
k_fs=floor(fc/fs);
kfs=k_fs*fs;

This part of code used to get information about spectrum.
- fc1 is the frequency of signal lobe that must be present in the interval [0 fs/2].
- kfs the beginning of this interval[k*fs (k+1)*fs].

This book can help you : Digital modulation techniques -Fuqin Xiong

regards,

Chaker

 

[suribright]

Hi,
Very good explanation and reference. In your code, I think one small mistake is s=[s A*cos(2*pi*fc*t+phi(i))]; not s=[s A*sin(2*pi*fc*t+phi(i))]; This is QPSK formula I think. I will rewrite my code as your recommend.
1. verify bandpass sampling. By adding the bandwidth and specify the interval

2. simulate with only one symbol : specify [k*fs (k+1)*fs]. and show FFT
Tks

 

[ChakerQ]

Hi,

yes i always confuse sine and cosine ^^'

I do this simulation before and i get the sine lobe at fc.

I hope our discussion helped you.

Good luck