This thesis aims to address the problem of how to develop an efficient traffic model that allows the prediction of the Session Initiation Protocol (SIP) server load in IP Multimedia Subsystem (IMS). By utilizing the characteristics of SIP messages, a signaling flow based approach to the traffic model is proposed. This model is based on an in-depth signaling flow analysis of a number of SIP session procedures defined in IMS and on the quantification of the SIP signaling traffic at the signaling flow level. The proposed traffic model allows the load of servers to be predicted with a simple mathematical calculation. Moreover, we develop a measurements architecture based on Event State Publication (ESP) and Event Notification Framework (ENF) in order to collect the network statistics and user behavior. According to the simulations that we carried out using OPNET, the model we proposed is shown to be acceptable. IMS network deployment cost optimization issue is formulated as a linear programming problem with various mapping strategies. |