At first view I see that you are using FR4 substrate with permitivity 4.3 and loss tangent 0.025, when in the IEEE document they use permitivity 4.4 and loss tangent 0.018.
May not be a big difference, but your results are not very far from the paper results. On S11 plots you have a bit of frequency shift and less return loss. With one exception (b), you have always S11 better than -10dB.
Otherwise I am not a big fan using PIN diodes to switch antenna elements to get multi-band operation.