Saturation magnetization parameter Ms should be independent of the interaction strength and sign (as well as on the particle size if the surface effects are excluded from the consideration). Real experiment conditions are always limited, so one measures a quantity only approaching to Ms, as Christian wrote. Therefore one should chose a proper model (for example, as it was proposed by Rasbindu, Langevin's superparamagnetic behavior thoroughly studied by C.P. Bean) to extrapolate experimentally measured dependence to extract true value of Ms parameter. Very often one should also take into account temperature dependence of the Ms and magnetic anisotropy parameters.

From general point of view antiferromagnetic interaction should decrease the ensemble magnetization value measured at finite magnetic field and temperature, while ferromagnetic coupling will increase it (not exceeding a true Ms value, however). 

In short, it can't if we really talk about the saturation magnetization. Magnetization is the average of the sum of the magnetic moments per volume. Saturation is achieved when each moment has its larges projection on the quantization axis and they are all aligned parallel along the quantization axis. The saturation magnetization is therefore independent of interaction. One has to keep in mind, however, that at finite temperature, saturation can only be reached asymptotically in the limit of very large applied magnetic field. 

For non-interacting nanomagnetic particles Langevin's theory is extended for super paramagnetic particles. If inter-particle /dipole-dipole interaction is included then one has to include their effect. Thus saturation magnetization calculated from the former will differ from the later. Best way to check by plotting M vs. 1/H and extrapolated to zero.

Saturation magnetization parameter Ms should be independent of the interaction strength and sign (as well as on the particle size if the surface effects are excluded from the consideration). Real experiment conditions are always limited, so one measures a quantity only approaching to Ms, as Christian wrote. Therefore one should chose a proper model (for example, as it was proposed by Rasbindu, Langevin's superparamagnetic behavior thoroughly studied by C.P. Bean) to extrapolate experimentally measured dependence to extract true value of Ms parameter. Very often one should also take into account temperature dependence of the Ms and magnetic anisotropy parameters.

From general point of view antiferromagnetic interaction should decrease the ensemble magnetization value measured at finite magnetic field and temperature, while ferromagnetic coupling will increase it (not exceeding a true Ms value, however). 

I agree with the earlier stated opinions of our colleagues: in General, the interaction should not affect the maximum possible value of the magnetization saturation. However, if closely spaced particles are in some kind of ordered structure (rigid lattice), then this could be cause of anisotropy, and then the magnetization curve will depend on the spatial orientation of the external field relative directions in this lattice. You should also consider the shape of the object containing the ensemble of particles, and the resulting magnetostatic energy. Therefore, for approaching saturation in the experiment, it should be necessary to apply fields of different strength, and depending on your setup, the actual values will vary slightly.

I think Saturation magnetization is not affected only the corecivity is reduces due to interparticle dipolar interaction as the  the result of strong exchange bias.

Please refer this article..

Beating the superparamagnetic limit with exchange bias

Vassil Skumryev*, Stoyan Stoyanov*, Yong Zhang*,

George Hadjipanayis*, Dominique Givord† & Josep Nogue

Nature

NATURE | VOL 423 | 19 JUNE 2003 | www.nature.com/nature

First of all I want to thank all you for your answers. @Nilesh thank you for the paper. I think not all the time dipolar interaction reduces the coercivity.  Alignment of easy magnetic axes of dipolarly interacted magnetic nanoparticles can show higher coercivity than their individual counterpart.  

whenever magnetic nanoparticle forms chain like structure they always aline the minimum energy conflagration i.e.  Alignment of particles along easy magnetic axes