Hwo can i design like this ?

#Matlab

if i have a matrix 6*6

which gave me info about the distance between all AP

I need to make 3 groups each group consists of 2 AP where AP not repeated in each group depend on the distance

Ex: i will start from AP no. 1 if i select it with AP no.2 as the dist. between the is large

so the second step i will check the 2nd group starting from AP3 & so on

CAN ANY one hwo can i design it by matlab code

Thanks all

[Walter Roberson]

Please say more about how you choose which two AP to group together ?

[sara alaraby]

i need to start from Ap no 1 chosse Ap depend on the longest disthemance between

[Walter Roberson]

Is the rule that at any one point, you select the lowest numbered unpaired AP as the start, and find the unpaired AP that is furthest away to pair with it?

I ask because the order you go through the APs matters. AP 2 might be the furthest AP from AP 1, but AP 1 might not be the furthest AP from AP 2.

[sara alaraby]

yes, this is my rule

[sara alaraby]

yes dear this is my rule

[Image Analyst]

sara, youi keep forgetting to attach your data, thus delaying a solution.

Answer 1

[Walter Roberson]

td = YourDistanceMatrix;
pairidx = 0;
pairs = [];
while nnz(td(:) ~= 0 & ~isnan(td(:))) >= 2
  std = size(td);
  [maxd, maxidx] = max(td(:));
  [maxr, maxc] = ind2sub(std, maxidx);
  pairidx = pairidx + 1;
  pairs(pairidx, :) = [maxr, maxc];
  td(maxr, maxc) = nan;
  td(maxc, maxr) = nan;
end

In the above code, distance entries can be nan for forbidden connections, and distances of 0 are ignored under the assumption that they indicate a point connecting to itself (which is not always true: in theory two AP could be a the same location, and this code will refuse to pair them together even if they are the "furthest" remaining APs.)

Because of the possibility of nans and extra 0s, the number of rows output in pairs will not necessarily be the same as half of the number of access points.

[sara alaraby]

Dear There is an error in line :

[maxr, maxc] = sub2ind(std, maxidx);

Any support please

[Walter Roberson]

[maxr, maxc] = ind2sub(std, maxidx);

[sara alaraby]

Dear

The pairs result on your code is matrix 20*2 , this not acceptable

I need to make only 3 groups so pair will be matrix 3*2 as AP not repeated,

can you help me to write the code like this ?

Thanks laot

[Walter Roberson]

td = YourDistanceMatrix;
pairidx = 0;
pairs = [];
while nnz(td(:) ~= 0 & ~isnan(td(:))) >= 2
  std = size(td);
  [maxd, maxidx] = max(td(:));
  [maxr, maxc] = ind2sub(std, maxidx);
  pairidx = pairidx + 1;
  pairs(pairidx, :) = [maxr, maxc];
  td([maxr maxc], :) = nan;
  td(:, [maxr maxc]) = nan;
end

[sara alaraby]

Dear ;

Can you help me ihwo can i update in this code if i have a matrix 15*15 a need to create 5 groups each one consists of 3 AP where AP not repeated

Thanks

[Walter Roberson]

You need to define the rule for grouping. You cannot simply extend to "three closest": A might be closest to B, and second closest to C, but that does not mean that B is close to C

           B+++++A++++++C
                 +
                 +
                 +
                 +
                 +
                 +
                 D

D is closer to B or C than B is close to C.

[sara alaraby]

Dear;

i need to start from AP no 1 then choose 2 AP which are fartest from it , in case of we select AP (5,10)

the second step we will start from AP 2 then chosse 2 AP with it and so son,

where AP not repeated

can you support to write like this

[Walter Roberson]

Note that the two AP that are furthest from the first node, could be very close to each other. That means that the order you go through is important. Are you sure that is what you want?

[Walter Roberson]

After you have done

[maxr, maxc] = ind2sub(std, maxidx);

then look for the maximum in td in row maxr but that is not located at (maxr, maxc) (so, the second greatest distance involving that same row.) Find its column and record that as part of pairs. Make sure you nan the entire corresponding row and column.

[sara alaraby]

dear i can't understand well what you mean

can you write such a code for me please

[Walter Roberson]

td = YourDistanceMatrix;
tripidx = 0;
triples = [];
std = size(td);
while nnz(td(:) ~= 0 & ~isnan(td(:))) >= 2
  [~, maxidx] = max(td(:));
  [maxr, maxc] = ind2sub(std, maxidx);
  td(maxr, maxc) = nan;
  [~, maxc2] = max(td(maxr, :));
  tripidx = tripidx + 1;
  triples(tripidx, :) = [maxr, maxc, maxc2];
  td([maxr maxc maxc2], :) = nan;
  td(:, [maxr maxc maxc2]) = nan;
end

I doubt this is what you want, but it is what you asked for and confirmed that you wanted.

I suspect that what you want is closer to finding the set of three that together cover the greatest triangular area.

Answer 2

[sara alaraby]

Dear;

yes dear i'm sure that this what i want

can you support me please

thanks