I used HFSS to simulate. And I got a result in another model, which achieves 1.1~2GHz(58%BW, 2.0VSWR), and the dimensions are the same as the one in 'Ultrabroadband Printed (UBP) Antenna'. The structure is similar to the one mentioned in the 'Design of wide-band aperture-stacked patch microstrip antennas', using Dual-offset feedlines. My ASP antenna (0.6~1.2GHz, VSWR<2) uses air substrate instead of foam-subs and Dual-offset feedlines as well. Now I achieved three peak frequency points from 0.6~1.2GHz, but the BW is still not that large. So I think if I adjust a better dimension, the results may be achieved. Thus, I try to change the dimensions of two patches, and then adjust the heights between the ground and patch1, and the heights between patch1 and patch2.