Do I need matching circuit for this antenna case?

[huytergan]

I designed and simulated a clover shaped cross dipole antenna with two port. (Two coaxial cable- 50 Ohms)

Design is like I provided below. I'm actually getting a good return loss within a relatively broadband frequency range.

But when I deactivated one of the ports to see Z matrix, real and imaginary part oscillate like this I provided below. And I wanna ask if I need a matching circuit although I have the bandwidth I desired when two coax used simultaneously.

Also info: I see many design research papers like that, they use two ports for the aim of circular polarization, but they never mention about a matching circuit. Is structural design enough to match impedance in this case?

 

[ferdem]

If you observe a good VSWR in EM simulation, you are done. Because you have a coupled structure, you have even and odd mod impedances. Driving the ports with 90 degree shift is neither even nor odd mode, I do not know whether a analytical description is available, EM solution may be the only way. For the coupled structures, excitation scenario may significantly change the results, you need to provide the same excitation scenario that you intent to use while simulating the structure.
Are the conventional z-parameters applicable for a coupled structure? Even and odd mode cases must be considered to find correct port parameters.