My material ( K0.5Na0.5NbO3) doesn't have any jcpds file format. For pure material itself researchers put different hkl value for the same composition. how to index it with the correct format?
Nita Dragoe gave essential general information concerning indexing program. In your case, there may be another route to solve your problem. Yes, it seems that your phase might be a perovskite, but depending on composition several crystal systems may occur. Your composition suggests that you are working with a 50/50 solid solution between KNbO3 and NaNbO3. Find out what crystal systems these two belong to. If they are isostructural, Vegard's law is likely to be obeyed, i.e. the cell volume (or any particular axis) is linearly related to the composition. If that be the case, it is quite straightforward to index your pattern accordingly.
I do not immediately understand what you mean by saying that different researchers index the same composition differently. You do not index a composition, you index the reflections of a certain phase. If different phases occur for the very same composition we are dealing with polymorphism, and in such a case the indices simply have to be different! That may be the explanation, but for high symmetry, indexing is fairly easy to make by hand.
You can find the mathematical relations between indices and cell constants for any crystal system in ordinary textbooks concerning diffraction. If the symmetry is cubic the relations are dead simple and you need only one parameter to solve your problem.
Study further, and you will most certainly arrive at the appropriate answer.