In the attached link you will find a really nice implementation of SVM for many languages, included MATLAB.
You also have some tutorials on this site, as well as many examples. If your "problem" is how to use SVM or any machine learning technique to be applied to your specific classification task, I would not spend time in developing your SVM solution, since there are nice implementations provided by many authors. However, if you are interested in going deeper in the SVM optimization, and try to improve some of the internal steps during the classification task, you can take the code as starting point.