FPGA Central

Hello, guys,

I am working on the 2X1 MIMO system simution, using simple Alamouti schem

and MATLAB. Now, i am adding sqrt Raised Cosine filter in the transmitte

and Receiver.
In the Tx, Modulaor(QPSK)---->Coding----->Pulse Shaping Filter,
if I have the symol sequence with length of 500, and the filter order is
40, after PSF, i have the new sequence with length of 540, then they
transmit in the Rayleigh fading channel.
Can i filter the data just after QPSK mod? or I have to get tw
transmitting sequences for TX1 and TX2, TX1=[x1 -x2* ...], TX2=[x2 x1*
...], then do the filtering? and can I filter the symbol directly? i mea
just use one PSF to filter I+j*Q, not one for I, another for Q.

OK, let's go to channel. I think the channel matrix is a 540 by 2 matrix


In the Rx, Filter----->Decoding----->Demodulator
I have only one receiver,
for time t, r1=h1x1+h2x2+n1,
for time t+T, r2=-h1x2*+h2x1*+n2,
then, i filter r1,r2 respectively? seems like i should filter the sequenc
r=[r1 r2 ...]?

after filtering, the output length changes. but the estimated symbo
depends on that ~x1=h1*r1+h2r2*
~x2=h2*r1-h1r2*
where h1, h2 is the cofficient of channel matrix, and have length of 540,
r1,r2 are the filtered received signal. Of coourse, they have differen
size, and can not be multiplied. So, this is the problem

there must be some logic mistake in my mind, but I can not find it...

I want to know to how to get the original data after filtering since th
squence length changes, and how to deal with the change in the receive
side.

I found something in the matlab help,
%% Filter Definition
% Define filter-related parameters.
filtorder = 40; % Filter order
delay = filtorder/(nsamp*2); % Group delay (# of input samples)
rolloff = 0.25; % Rolloff factor
Tx:
% Upsample and apply square root raised cosine filter.
ytx = rcosflt(y,1,nsamp,'filter',rrcfilter);
Rx:
% Filter received signal using square root raised cosine filter.
yrx = rcosflt(ynoisy,1,nsamp,'Fs/filter',rrcfilter);
yrx = downsample(yrx,nsamp); % Downsample.
yrx = yrx(2*delay+1:end-2*delay); % Account for delay.

I used this, but the above problem still exists.
REALLY NEED YOUR HELP! THANK U GUYS IN ADVANCE!



_____________________________________
Do you know a company who employs DSP engineers?
Is it already listed at http://dsprelated.com/employers.php ?

On May 24, 8:40 am, "cellarway" <cellar...@gmail.com> wrote:
> Hello, guys,
>
> I am working on the 2X1 MIMO system simution, using simple Alamouti scheme
>
> and MATLAB. Now, i am adding sqrt Raised Cosine filter in the transmitter
>
> and Receiver.
> In the Tx, Modulaor(QPSK)---->Coding----->Pulse Shaping Filter,
> if I have the symol sequence with length of 500, and the filter order is
> 40, after PSF, i have the new sequence with length of 540, then they
> transmit in the Rayleigh fading channel.
> Can i filter the data just after QPSK mod? or I have to get two
> transmitting sequences for TX1 and TX2, TX1=[x1 -x2* ...], TX2=[x2 x1*
> ...], then do the filtering? and can I filter the symbol directly? i mean
> just use one PSF to filter I+j*Q, not one for I, another for Q.
>
> OK, let's go to channel. I think the channel matrix is a 540 by 2 matrix.
>

Honestly I can't read all of your post, it's very hard to follow.
But what makes you think that the channel is 540 (time) by 2 (whatever
this dimension is)?

In the Alamouti scheme the channel is assumed to be:
1) Flat fading.
2) Known by the receiver, and does not change for at least one
space-time symbol.

So I think that your way of generating your channel is wrong.

Julius