In urban cities, many MUs (mobile users) follow a fixed mobility pattern. Most of the time, these users can be found in a particular registration area (RA) at a certain time period. We say that MUs follow a fixed set of RAs on a daily basis. Therefore, an MU can be located at one of the RA from the set of fixed RAs when a call arrives. In prior schemes, the location update cost is minimized by increasing paging cost. In this paper, we propose a mobility agent (MA) and adaptive paging strategy based on MUs' movement. The MA is a special functional entity in the network, and many RAs are connected to MA as per mobility pattern of an MU. The paging area is selected based on the average speed of MUs. The major features of the proposed scheme are the overall reduced location management cost and less paging time compared with existing scheme. We present analytical modeling and performance evaluation that show that the proposed scheme has less location management cost and less paging time to locate an MU compared with existing scheme.