question about methods

What mathematical methods of analysis of broadband microstrip antennas provide the best exactness in 3-10 GHz bandwidth? A have read a lot about MPIE(mixed potential integral equation) method, but it seems to be useful only for patch antennas (with a ground plane on a substrate)

A time-domain technique is probably best for broad band results, especially when there are a only a few (maybe just one) ports, as is likely the case with your antenna. Time domain techniques solve for the impulse response and then transform the impulse response into the frequency domain yielding broad band S-parameters. I don't know if there are any planar time-domain tools (does anyone know of any? If so, please post!). However there are many volume meshing tools that work in the time domain. I am most familar with CST that uses Finite Integration Technique. There are also the closely related Finite Difference Time Domain (FDTD), and Transmission Line Method (TLM). Lot's of theory published. Any of those three techniques will give good results.

CST Microwave Studio is really good program for RF simulation (I have used it to construct different sstructures). But for my master thesis i need to construct a mathematical model of an ultrawideband dipole antenna. For the fast computing it is desired to make a model on the base of integral equations, but I have not seen any articles about this problem (only a microstrip PATCH antenna is good described). Do you know any good literature about this problem?