OFDM - Frequency Selective Fading

frequency selective fading ofdm

Does any one know how to implement Frequency Selective Fading in OFDM ??

frequency selective fading matlab

I think you can Rayleigh channel fading to implement it......The function is available in MATLAB...

ofdm frequency selective matlab

hi Barath...

I appreciate ur answer. I wanted to know wrt ODFM specifically.

As you know, OFDM would convert the modulated signal into a time signal with IFFT operation and do the inverse at the receiver.

We can calculate the Fading Coeff. But how do we find the CIR and how do we do convolution with the OFDM signal ?

Im slightly confused... So please highlight..

ofdm frequency selective fading

yesh_r1,

I am sorry I am an amature in OFDM I just created a rayleigh filter using that function and passed the data bits on to it before taking the fft at the receiver..
And that would implement frequency selective fading..

Thought that I can share this with you as I did that only.

frequency selective fading and ofdm

hey buddy..i am an amature too... y sayin sorry n all..

Thnks for ur help..

I will chk out the function which u are referring to ..

fading ofdm in matlab

I guess in the simplest way....you can mutiply each of your subcarrier with a flat fading coeeficient.....

ofdm-frequency selective fading

hi

you can use the matlab's rayleighchan and ricianchan functions

flat fading convolution

the behavior of the channel is generally specific to the radio environment. the frequency selectivity of the channel only depends on the system bandwidth. in other words, given the same channel (environment), it can be either frequency-selective (if the system bandwidth is large) or frequency-flat (if the system bandwidth is narrow).

Therefore, in order to simulate the frequency-selective channel, you just need to simulate multiple taps with the maximum delay larger than the sampling time.

frequency selective fading model pdf

Hi.. Thanks for ur response..

I am actually lookin at Freq Selective fading... But I have an elementary question.

Assuming I have an array of mapped data (QPSK) 2x256 ..

After using IFFT operation I would now have OFDM data sized [1x640] ..

I have found out the channel coefficients seperately..

How do I perform Convolution operation between the two(OFDM and Channel Coeff) ?
(Coz, the ch coeff are supposed to change for every time period greater than Nyquist time...)

(I hope I have explained my question clearly...

ofdm fading

You cannot use the Matlab conv here but if you really understand the convolution technique here's what you can do.

1. pre-pend and append n-1 zeros to your post-ifft ofdm signal. n = channel length
2. form an n-length vector from the zero-appended ofdm signal.
3. start a loop for m iterations. m= n+k-1. k=length of raw ofdm signal
4. for every iteration i -
output(i)= ofdm(i:i+n)*transpose(channel(i,)
end

now I am assuming that ofdm signal and channel(i, are both row-vectors
hope this helps

Added after 1 minutes:

ok the editor screwed up " : ) " and turned it into smileys

frequency-selective channels matlab

hey.. that was a nice suggestion .. thanks..

I was trying to use the SUI channel model for the simulation...
I have attached a file ChannelModels.pdf which shows how fading coeffs in SUI channel models are calculated....

The following were the steps I found in an online source( not sure if it works )

------------------------------------------------------------------

1.) fading_coeff in an array [tap size X Number of OFDM frames ]
Refer attachment : channelmodels.pdf (page 33)

2.) use the delay spread and OFDM Symbol time to determine the convolution
locations
Tb = Useful symbol time

ofdm sample time = Tb/256;

convolution Locations = 1 + round( tau / ofdm sample time ) ;

3.) Assuming 'cir' which stores the Channel Impulse response;

cir = zeros(1,max(convolution Locations));

for i=1:length(sim_param.Ch.tau)

cir(sim_param.Ch.conv_locs(i) )=cir(sim_param.Ch.conv_locs(i) )
+ fad_coeff(i,counter) ;

end

4.) Finally convolve cir and rx signal

----------------------------------------------------------------

do u think this would work ?


Thanks....