In this paper we develop an approach of embedded Markov chain to analyze the signaling cost of movement-based location update (MBLU) scheme implemented in a personal communication service (PCS) network with location area (LA) architecture. We propose to use a fluid flow model to characterize the dependency between cell residence time and LA residence time. We derive analytical formulas for the expected numbers of various types of location updates and for the expected paging area size. The accuracy of the analytical formulas is tested using a simulation. The analytical formulas can guide the implementation of the MBLU scheme in PCS networks. Moreover, the modeling approach can be explored to study other LU schemes.