Hi H B,
The short answer to your question is that there is not an obvious reason which leads to some differencies in the results.
The trend of your model's S11 curve looks close to the one in the paper. Since you can't be sure that you are simulating exactly the same model with the one in the paper (same boundaries, same ports) some differencies in the plots should be expected.
To get closer results for the lower end of the frequencies, I would suggest to do the following:
1) Push the convergence criteria to a lower number and enforce a denser mesh.
2) Use a discrete frequency sweep with sufficient number of points for the lower end of the Frequency range.
You are navigating away from the AIS Discovery experience