Normalizing farfields for qualitative comparison...

Hey folks,

i finished my 2D farfield algorithm for user-defined structures and want to compare my results with others. What is the best way normalizing farfields ? Directivity ? Power ?

Regards

Winn