A dynamic distributed overload control mechanism in SIP networks with holonic multi-agent systems

Nowadays, session initiation protocol (SIP) is the most important application layer protocol for multi-media to create, modify, and terminate the sessions. SIP is the platform of next generation networks. In this way, SIP should be able to respond to the needs of such a largely-used network. One of the major problems in SIP networks is overload. The basic mechanism embedded in SIP cannot appropriately resolve the due problem by purely rejecting requests, which may cause collapse in network. Many methods have been proposed to overcome overload in SIP, among which, multi-agent systems are new agent-based and increasingly growing approaches. Multi-agent system is a powerful tool to model and develop the complex large scale distributed systems. In SIP network, each agent may point out to a server which is in interaction with the others. As a result, a distributed and complex system is simply modeled. The due complexity can be reduced by holonic organization. In this paper, SIP network is divided into geographical areas in which each holon controls an area. The entire network is controlled in hierarchical structure of holons. Holonics structure can be dynamically changed based on the circumstances parameters and the status of the network during execution. Holons implement an end-to-end window-based method. This method offers an appropriate window size for edge servers to control load from the beginning of the network and prevent network overload. The simulation results confirm improved throughput, as well as better reduced call setup delay, rapid response to change and fairness.