But if you want to strictly prove p!=np, you should rely on some kind of random number generation algorithm. or maybe i haven't fully understood it.
He random number generator part has already been discussed earlier. The one-way function he presented is a good step towards a proof, it can possibly be used as a basis for a formal theorem of p!=np but I don't see any such theorem in the paper.
I can't wait to hear about the use cases for this encryption scheme.