I am confused whether each magnetic atom contribute to the magnetization direction or resultant of magnetic atom in an unit cell form contribute to magnetization of that make spin wave.
Generally of the atomic spin, however one in principle may also deal with a coarse-grained model where one deals with resultant spins. Operators for quantum spins form a Lie algebra, however at the level of spin-wave approximation, the problem is essentially no different from that of, say, phonons (which are bosons - as opposed to hard-core bosons).