Polling system with threshold control for modeling of SIP server under overload

Abstract

The main purpose of this research is the development of the methods for realizing an overload control mechanism on the Session Initiation Protocol servers by the application of polling systems with different service disciplines. The mathematical model is studied by means of numerical methods of the queuing theory and allows analyzing the behavior of different control parameters depending on the load in network of SIP servers. The polling system consists of two queues of finite capacity and implements the threshold control of loading by low-priority customers. The exhaustive and gated service disciplines are studied under Markov assumptions; formulas for calculation of the main probability measures of the polling system are derived. By performing simulations we demonstrate that the polling system with a threshold in the priority queue is a possible solution for loss-based overload control scheme at the SIP server. In some cases from the viewpoint of server utilization we found that the gated discipline is more preferable.