Managing users' mobility in access networks remains a challenging task for the next generation cellular networks, especially with the increasing mobile data traffic and the variety of users' consumption behaviours. Cloud Radio Access Network (C-RAN) is a newly introduced radio access network that provides a promising degree of flexibility in the management of cellular networks, thanks to its centralized architecture. In our work, we focus on the optimization of the dynamic C-RAN structure. We propose a methodology that allows to adapt the C-RAN topology to real users' consumptions and their mobility patterns. We test our strategy over a real-world dataset providing usage patterns of an operator's customers in an urban area. Our results indicate that our algorithm allows to reduce the signalling overhead resulting from the users' mobility.