Replacing the non zero values of matrix by another values based on some criterion.

Hello,

I have a channel matrix similar to the given matrix.

A=[1 0 0 0; 0 0 12 0; 0 2 0 0; 8 0 0 0; 0 9 0 0; 0 0 0 8; 4 0 0 0; 0 0 0 3; 0 5 0 0; 0 0 6 0]

I have to find the non-zero element (channel gain) in each column and multiply them by power factor in ascending order of their channel gains. And update the matrix. The value of power factor is calculated as

2*i/(K*(K+1)). Where k is total number of non-zero elements in each column and the value of I varies as: i=1,2,,3….K. . For example, in column 1 there are three non-zero element (1, 8, 4). Then power allocation is as follows:

Ch gain: 8>4>1

Power factor: 2*1/(3*4), 2*2/(3*4), 2*3/(3*4) for 8, 4,1 respectively.

Problem: I am able to find first maximum of each column and able to update its value but not able to do it for other. Even in the final matrix only the first max value of column is updated. If anyone have idea please help me. The raw structure of code is given here.

A=[1 0 0 0; 0 0 12 0; 0 2 0 0; 8 0 0 0; 0 9 0 0; 0 0 0 8; 4 0 0 0; 0 0 0 3; 0 5 0 0; 0  0 6 0]
for j=1:1:4
Totl_users_to_each_BS(:,j)=sum(A(:,j)~=0)
end
T= Totl_users_to_each_BS;
%%%%% power allocation factor values
for j=1:1:4
for k=1:1:T(1,j)
%NAV= 2.*k.*A(i,j)./(T(1,j).*(T(1,j)+1))
NAV= 2.*k./(T(1,j).*(T(1,j)+1))
end
end

for j=1:1:4
[A1 I] =max (A(:,j))
%[A2 I2]= max (A(:,j))
A3=A(I).*5%%% we need to take NAV here
A(I)=A3
end

[Adam]

doc sort

should help, to just give you your column sorted and then apply the power values to the whole column with a .* multiplication.

[Shweta Rajoria]

Thank you Adam... On applying sort power allocation can be performed but the updated matrix have different index value and that is not accpectable.

[Adam]

You can return the indices as the second argument from the sort function and put them back in their original order afterwards.

Answer 1

[Guillaume]

If I understood correctly,

A = [1 0 0 0; 0 0 12 0; 0 2 0 0; 8 0 0 0; 0 9 0 0; 0 0 0 8; 4 0 0 0; 0 0 0 3; 0 5 0 0; 0  0 6 0];

[sortedA, roworder] = sort(A, 1, 'descend');
K = sum(sortedA ~= 0, 1);
powerfactor =  2 * (1:size(A, 1))' ./ (K .* (K+1));  %your 2*i/(K*(K+1))
Afactor = zeros(size(A));
Afactor(sub2ind(size(A), roworder, repmat(1:size(A, 2), size(A, 1), 1))) = powerfactor; %reorder powerfactor according to original order of A

result = A .* Afactor

[Shweta Rajoria]

Thank you Guillaume..

It seems to a little bit clouser. Thank you so much. But, power factor should be calculated for non-zero element for each column in A. For example in column 1, there 3 non-zero elements. So, in power factor calculation, i values are i=1,2,3.

Simillar for column 3, there are 2 non-zero elements. So, power factor is calculated for i=1,2.

Here, range of i is changing for each column as per the element value in K=[3 3 2 2].

And sir, there is error in power factor line while executing your code.

If you have any idea please help me.

[Guillaume]

It doesn't matter if powerfactor is calculated for i > K since these values will be multiplied by 0. You get the correct values anyway for i <= K.

Since you didn't specify a version of matlab (up there on the right of the page, there's a field for you to enter your version), I assume you were on a more or less up to date version. If you're on a very old version (< R2016b), then yes the powerfactor line will error, in that case:

powerfactor = 2 * bsxfun(@rdivide, (1:size(A, 1))', K .* (K+1)); %for versions <R2016b

[Shweta Rajoria]

Thank you sir. It works.