Self-organization of hierarchy of system has been focused in task allocation of distributed autonomous systems and network analysis. It is important to realize the mechanism of hierarchy generation for implementation in artificial systems. In order to know the principle, we try to model the control of caste differentiations in the termite ecology. Equations of evolution are created, using both of biological data and assumptions obtained by mathematical analysis. In addition, the model is validated by computer simulations. In this study, we propose that the probability migration of individuals and modulations of fluctuation are operated as a differentiation control strategy.
