feed for horn using waveguide port or discrete port
I know in HFSS it would be a wave port. HFSS has the terms "lumped port" and "wave port" and I suspect the waveguide port is what you want in CST. A horn is going to be connected to a waveguide, so if CST has a waveguide port, I'd be very surprised if that was not the one to use.
As for getting S11, I don't know, but this sounds the sort of thing that would be in a tutorial. Have you done the examples in the CST manual?
There seems to be a large number of people asking questions about HFSS, which I know are covered in the first couple of examples of the manual. I don't know why people don't RTFM.
Dave
u can find an example for this in CST folder
Waveport solve a 2D eigenvalue problem first and find out all the possible mode that can be supported by the waveguide, so waveport is very accurate in calculating S-parameters b/c the excitation modes are completely going into the waveguide without any reflection. However, a lumped port can also form your excitation, but you will have a lot of evanescent mode going into the structure at the beginning. If you measure your field at a deep position in the waveguide, waveport and lumped port generate about the same waveform, however, to calculate S-parameter, waveport is more accurate. In reality, waveport matches the case where a waveguide is already connected to a mode-matched excitation, this, however, in most cases is not valid, because to excite a real waveguide you always use a pin from coaxial cable penetrate inside it. which excite all modes and let waveguide filter out the supported mode.