Lease Square Minimization, lsqnonlin

I am using lsqnonlin for LSM problem. I have tried both ways trust region reflective and levenberg marquart (can be changed from options). They both provided the same result. What are the difference in these methods? PS: I have 4 equations with 3 unknowns.

Answer 1

[Svein Olav Hegerland]

Hi! I know it a long time ago you posted this post, but can I have a look at your code :)? I am trilaterating using the lsqnonlin solver and would like to compare my approach to yours.

[Rafa Z]

check your FB;

[Hussein Kwasme]

Hello, Just found this discussion, I'm trying to do trilateration of a point in a 2-D plane having 4 reference points (known positions and distances to my unknown point).

Is it possible to have the code? I can solve my system with SOLVE command if I have only 3 reference points (shown in picture), but I get no solution when I add the 4th reference point.

if true
BS=3
  syms x1 y1 m
%     assume(x1,'real')
%     assume(y1,'real')
  for i = 1:2:BS
      syms x1 y1        
      %solving the equations
      m(i,:) = (((x1-anchorLoc(2,1)).^2 + (y1-anchorLoc(2,2)).^2 - (distanceNoisy(2))^2)) - ...
          (((x1-anchorLoc(i,1)).^2 + (y1-anchorLoc(i,2)).^2 - (distanceNoisy(i))^2));
  end
  syms x1 y1   
  [X1,Y1] = solve(m,x1,y1);
  beacon_est(1,1)=double(X1);
  beacon_est(1,2)=double(Y1);
end

[Rafa Z]

shoot me a message to [e-mail]; will share scripts I have.