Hi,
See the attached file i have corrected it. The first Notch is exactly at the same frequency range as desired,the second is a little earlier from 5.01-onwards. The notches have relatively low VSWR because FR4 is a lossy material and above 1GHz losses of FR4 gets high and that is the reason 3rd Notch is not even appearing. Apart from this Feedline width is not correct for 50Ohms,it should be around 1.9mm to achieve 50Ohms characteristic impedance. How you have decided for 3mm width???? Any special reason???? Your Airbox size wasnt correct as Lamda/4 is 25mm not 12.5mm if you calculate Lamda using f=3Ghz your Lumped port excitation was defined for 36Ohms,i dont know why. I have corrected it for 50Ohms aswell. You have changed the substrate so microstrip feedline width ll be changed and accordingly all other dimensions will change in order to get the same exact results of the paper you have mentioned.
My suggestion: Use the same substrate as specified in Paper and try to replicate the results of the paper by using exact dimensions mentioned in paper. That is always the first step. Once it is done,then you can change the substrate to FR4 or any desired material and then see its effects on the result. In order to maintain the same results you can play with dimensions of different parts of antenna geometry. I hope you have got my point.
/SC