Perhaps you should contact the authors of the paper.

Did you check the simulation with SMA connector? 

Did you check the substrate properties with frequency change?

Did you check the accuracy convergence and also check the meshing of the model?

Dear Ali Al-Azza

1) I did check the simulation using wave port,  sorry I didn't check the simulation using SMA connector. With your help I may design it, Please provide the necessary steps to design SMA connector.

2) I believe I plot the substrate properties with frequency change, I attached the screenshot of it please correct me if am wrong.

3)I checked the accuracy of the convergence it is converged within 4 passes. I also checked the meshing of the model it is length based adaptive meshing,

I am not familiar with HFSS, In the paper,the authors used a FR4 substrate with dielectric constant of 4.4. From your plot er is changed from 2-2.18 along interesting BW.

Please, correct the definition of substrate dielectric constant. and try to design the SMA connector.

Last check, you can run the same design on CST simulator.

if the results if HFSS and CST is the same, Then you should contact the authors to ask them about the correct dimensions.

Thanks 

The frequencies you are aiming at, FR 4 is not right material. Even if analysis offers good results, actual to achieve is very difficult. I have not come across, so far, FR 4 practically works for such high frequency. Due to production limitation, it is non homogeneous media for the frequencies