The three suggested projects for any student developers looking to work on KWin include refactoring KWin's core, a unit testing framework for the X11 window manager, and initial support for Wayland clients.
Refactoring the KWin core is the most important GSoC KWin project in Martin's eyes as it's critical for the future Wayland support and other work. KWin has evolved from being an X11 window manager to being an OpenGL compositor with X11 window manager, but now its core must be reworked so that this existing code can still be leveraged when targeting Wayland clients.
The "initial support for Wayland clients" project is not to create a polished, working KWin-Wayland implementation, but rather more of a research project. Martin admits this code would likely not be merged to master. The project would explore adding Wayland clients to the compositor so they could be rendered like a normal X11 window. This work would benefit Martin and other KDE developers in understanding the Wayland support requirements. "To make it clear: a complete port of KWin to Wayland is not possible in the scope of a GSoC."
If there's any student developers interested in applying for this KWin-Wayland work, see Martin's blog post.
Speaking of KDE on Wayland, last week Nokia Labs talked about bringing their Qt Compositor up on Wayland.