Problem with implementation of Linear SVM with quadprog

I have to implement a classificator SVM (Support Vector Machine) in Matlab. The samples are the measurement of diabetes in the Pima Indians. Every sample have 9 features (8 + class):

1. Number of times pregnant
2. Plasma glucose concentration a 2 hours in an oral glucose tolerance test
3. Diastolic blood pressure (mm Hg)
4. Triceps skin fold thickness (mm)
5. 2-Hour serum insulin (mu U/ml)
6. Body mass index (weight in kg/(height in m)^2)
7. Diabetes pedigree function
8. Age (years)
9. Class variable (0 for not diabetes and 1 for diabetes)

In my program I follow this few steps:

1. Load data from data.txt (768 samples)
2. Delete all incomplete samples (392 samples)
3. Implementation of the SVM alghoritms (using quadprog)
4. Display results

The problem is that every time quadprog send me this message:

Maximum number of iterations exceeded; increase options.MaxIter.
To continue solving the problem with the current solution as the
starting point, set x0 = x before calling quadprog.

I tried to solve this problem in three ways:

1. increasing the number of iterations (currently is set at 1000, but I also tried higher values without results)
2. changing UB (if set to infinity the solution is unbounded)
3. implementing a PCA algorithm and reducing the number of features (the error launched by quadprog remains unchanged)

Since it is a work for the university, I have to implement a LINEAR SVM with quadprog (this is very very important). Can you help me? Thanks and sorry for my bad english

close all
clear all
clc
%
% load data
filename = 'data.txt';
vsample = importdata(filename,',');
vsample = vsample';
%
% delete incomplete data
vsample(:,vsample(2,:) == 0) = [];
vsample(:,vsample(3,:) == 0) = [];
vsample(:,vsample(4,:) == 0) = [];
vsample(:,vsample(5,:) == 0) = [];
vsample(:,vsample(6,:) == 0) = [];
%
% training data for svm
x = vsample(1:8,:)';        % samples matrix with 8 features
y = (vsample(9,:)'*2)-1;    % class array (+1 or -1)
%
% implementation of the LINEAR svm
svm.D = size(x,2);
svm.L = size(x,1);  
H = zeros(svm.L,svm.L);
for i=1:svm.L
    for j=1:svm.L
        H(i,j) = x(i,:)*x(j,:)'*y(i)*y(j);
    end
end  
f = -ones(svm.L,1);
A = [];
b = [];
Aeq = y';
beq = 0;
LB = zeros(svm.L,1);
UB = inf*ones(svm.L,1);
options = optimset('MaxIter',1000,'LargeScale','off');
alpha = quadprog(H,f,A,b,Aeq,beq,LB,UB,[],options);
svm.sv = find(alpha>0);
svm.ns = length(svm.sv);
svm.w = sum(((alpha.*y)*ones(1,svm.D)).*x)';
svm.b = y(svm.sv(1)) - sum((alpha.*y).*(x*x(svm.sv(1),:)'));
%
% testing with same samples
y_est = sign(x*svm.w+svm.b);
prc = (sum(y_est(:)'== y(:)')/svm.L)*100;
disp([fprintf('\n'), 'Percentage of samples correctly classified: ', num2str(round(prc)), '%']);