Indeed find some software that can do 2,5 D electromagnetic simulation and you can optimize the structure. The wider the stub at the end, the bigger the bandwidth but the length should be adjusted likewise. Therefore simulation is the best option.
I knew I had seen this recently in one of my books.
Take a look at Rainee Simons, "Coplanar Waveguide Circuits, Components, and Systems". Several variations are discussed. Wadell, "transmission Line Design Handbook" has a little more info.
VSWR is right ..... stfw. Lot's of references there.