Convintrlv: What is the best way to implement convintrlv for one packet?

openExample('comm/DigitalVideoBroadcastingTerrestrialExample','supportingFile','commdvbt.slx')

I have been using this example as a reference to build a DVBT simulation on Matlab with code rather than a simulink block diagram. One problem I have encountered is that I seem to have to zero pad the convolutional interleaver by a lot to be able to have enough data for the deinterelaver to recreate the original message.

simple example to show what I mean:

nrows = 3

slope = 1

x = [1 2 3 4 5 6 7 8 9 10];

x_prime = convintrlv(x,nrows,slope) = [1 0 0 4 2 0 7 5 3 10]

x = convdeintrlv(x_prime,nrows,slope) = [0 0 0 0 0 0 1 2 3 4]

Obviously it is decoding the message but theres a delay and Id have to introduce a delay to get the whole picture. So then I pad with zeros.

x = [x 0 0 0 0 0 0];

x_prime = convintrlv(x,nrows,slope) = [1 0 0 4 2 0 7 5 3 10 8 6 0 0 9 0]

x = convdeintrlv(x_prime,nrows,slope) = [0 0 0 0 0 0 1 2 3 4 5 6 7 8 9 10]

Cool, but now my message has turned from a length 10 packet to a length 16 packet and this will affect later subsystems. What is a good way to solve for this?

I know in the DVBT example I showed at the top there is a buffer command. I'm not sure how to implement it, though.

Answer 1

[Shivam Gothi]

Hello,

I see that you are working on implementing convintrlv and convdeintrlv using MATLAB code. However, it seems that the delay introduced in output signal and the zero-padding stages are not ideal for your requirements.

While implementing convintrlv and convdeintrlv, the delay is inhereritant and we cannot do any thing to remove.

But there is a work-around for this.

To implement it as a MATLAB code, you can automate the process of padding extra zeros to input signal and removing the excess zeros from the output signal by defining a "user defined function".

I have made two user defined functions in files "UserDefined_convintrlv.m" and "UserDefined_convdeintrlv.m". They are attached with the answer.

You can use them as shown below:

x = [1 2 3 4 5 6 7 8 9];
x_prime=UserDefined_convintrlv(x,3,1);

the output of above code is:

Now, again se the user defined function "UserDefined_convdeintrlv" as shown in below code:

 x=UserDefined_convdeintrlv(x_prime,3,1)

The output of above command is attached below:

In this way you can get rid of manuall adding the zeros. Also, the output generated by "convdeintrlv" is free from any delay.

I have tested both the user defined functions for different values of "nRows" and "slope". I found that it is working.

I am also attaching the code of "UserDefined_convintrlv.m" and "UserDefined_convdeintrlv.m", in case the attached files failed to open.

%User defined function for convintrlv. Save the file with name "UserDefined_convintrlv.m"
 function out = UserDefined_convintrlv(x,nRows,slope)
    zeros_to_be_padded=zeros(1,nRows*(nRows-1)*slope);
    x=[x zeros_to_be_padded];
    out = convintrlv(x,nRows,slope);
end

%User defined function for convdeintrlv. Save the file with name "UserDefined_convdeintrlv.m"
 function out = UserDefined_convdeintrlv(x,nRows,slope)
    zeros_to_be_removed=nRows*(nRows-1)*slope;
    temp_out = convdeintrlv(x,nRows,slope);
    out = temp_out(zeros_to_be_removed+1:end);
end

"REMEMBER TO PLACE THE "UserDefined_convintrlv.m" and "UserDefined_convdeintrlv.m" FILES IN THE SAME WORKING DIRECTORY.

I hope this helps...