If your material is pure and you have high quality data in terms of peak positions then you need to carry out 'indexing' of the pattern. This should give a solution showing the symmetry group and cell parameters plus a suggestion of what the space group could be. There is software available for this in the weblink below.
DICVOL, ITO and TREOR are popular and available in the Crysfire package. It's best to use more than one package and look for common solutions.
Hi Ian, in fact this procedure only enables a first guess about the Bravais lattice. Everything else (symmetry group or even space group) shouldn't be possible because of Friedel's law. This is of course different to single crystal techniques which enables a first guess about glide planes or screw axis. However, there also triclinic phases exist which are displayed by a cubic translation lattice. A gamma' powder of a Ni-base superalloy you would also identify as F lattice since the small 001 is practically invisible. In TEM you can see it and you know it is a P lattice. Therefore, I would better say, the indexing is mostly a first guess only...
Are there any hallmarks (hint)of the diffraction pattern of hexagonal and the cubic structure so that when we see the diffraction can distinguish directly ?