There are several methods used for this test. One is with a microstrip above an EBG ground. One is with a microstrip suspended in air above the EBG surface, which has its own dielectric backing. See for example "A uniplanar compact photonic-bandgap (UC-PBG) structure and its applications for microwave circuit".
The "proper" test will depend on whether or not your EBG has a conductor backing of its own.
i will show you what i try. i design micrstrip line in front of the substrate and EBG in the back for the same substrate.is this way is correct
also for analysing the badgap, the result base 0n s12 or s21
It's been done that way before, yes. I'm a little surprised by your S-parameters though, why are S21 and S12 so different? They almost appear to be opposites around 1.9 GHz... is "S21" actually S11 or S22?