Deterministic-Based Performance Modeling of a Cluster of Nodes Handling Subscriber Profile Query and Update in CDMA Mobile Switching Center

摘要

This paper presents performance modeling of processor occupancy in a cluster of nodes handling subscriber profile query and update in a code division multiple access (CDMA) mobile switching center (MSC). It also presents a simplified method to determine the profile update ratio that represents the fraction of autonomous registration (AR) and call events requiring subscriber profile updates. Four models are presented. The performance models developed offer a number of advantages, including enabling prediction of the home visitor location register (HVLR) capacity in a deterministic fashion to facilitate subscriber profile management and planning and to provide a simple method to determine the subscriber profile update ratio. The models are simple to use and provide accurate predictions. The models are platform-independent and can be easily applied to other platforms by substituting a new set of costs per message. Performance improvement by update bundling is also demonstrated, and future work on modeling update bundling is presented.