One possible approach would be to search for current uses for REST. Then get a simple RESTful Web service running on your own. Finally, think of a few ways you can either improve on the services you've investigated, or perhaps come up with an idea for a new area for using REST. Or maybe an improvement to the Oracle RS API. Or an extension to servlets to support all HTTP methods. Or [fill in the blank with a cool idea]...
For a master's degree, the innovation does not need to be earth-shattering. Also, some universities allow an alternative to the thesis in the form of a research project, so if you can't figure out an innovation, maybe you can figure out, say, a best practices commonality among implementations. Be creative. Give your advisor a few options, and maybe he/she will approve one of them.