PML is the same as the air box with radiation boundary with some few differences. First, PML in general makes it harder to reach convergence (requires more RAM). On the other hand, taking about advantages, PML absorbs a much wider range of waves in terms of frequency and direction, so PML can be closer to the discontinues in your model; and as a result, your model would be smaller. Finally, PML gives more accurate results (like VSWR of the antenna), but it really depends on your computer and your model size (frequency).

PML Setup Wizard:

Hope I could have answered your question. Let me know if you have any more questions.

Good luck!

Thank you sir for your clear explanation and interested in my question.

AFAIK (i.e. I may be wrong), putting PML close to a waveguide is a bad idea. It is perfect only in normal direction, only at designated frequency. But near/inside the waveguide anything can happen. It is not the same as putting PEC, PMC, or periodic boundary. Generally, the rules are the same as for standard radiation boundary (was it Mur's, or Engquist-Majda boundary?), just the performance is better. So, yes, PML can be placed closer. There is another reason, why: figuratively, PML works like a sheet absorber, not just a load. It also absorbs evanescent waves, and can be applied to quasi-TE/M waves in near-fields. Simple radiation boundary in HFSS cannot handle this. But you still should not rely much on these properties of PML, because its performance is not universal.

As a result of my research, I learned that I should not use the PML Setup Wizard, because my design is a waveguide and it takes dual polarization on circular port and seperates as vertical and horizontal polarizations. So, I do not need a radiation boundary for my design, because my design does not radiate toward the space like an antenna or an absorber etc.