I think this is way to simple for a graduation project. It's more or less a one weekend task. Actually you can find it already implemented in c#: http://www.c-sharpcorner.com/uploadfile/b942f9/implementing-the-dbscan-algorithm-using-C-Sharp/
If you think in this kind of topic I suggest you better do a comparison of several clustering algorithms. To do that you have to choose sample field (any data source that can be assessed with this method, and the resulting clusters can be compared in some form). You could assess the efficience of the algorithms in theory (evaluating complexity), the performance in you implementation on several different sample sets (differing in content and size), and maybe performance improvement considerations too. At the end you could compare the resulting clusters that were delivered by the different clustering methods on the same sample.
Well this might be enough for a graduation project. At least where I came from...