Research On Key Technologies Of Overload Control For Ip Multimedia Subsystem

Overload occurs in the network when the network has insufficient resources to handle offered load. Under overload, the throughput of the network can drop significantly, which results in the bad impact on the quality of service. Thus it is crucial to equip the network with the overload control mechanism. As the evolution of the core of the next generation network, the IMS (IP Multimedia Subsystem) implements the separation between the service control and the call control as well as the separation between the call control and the bearer control. In order to provide better support for the multimedia services, IMS takes charge of session control, charging, authorization, authentication, etc. Thus it can be regarded as the central nervous system of the network. Therefore, in order to guarantee the quality of service, it is very necessary to control overload for IMS.In this dissertation, the author’s major research on the IMS overload control is summarized as follows:the IMS overload control architecture is divided into the network and transport layer, the control and interconnection layer, and the service layer; as for the control and interconnection layer, a sender-based end-to-end SIP (Session Initiation Protocol) overload control policy is proposed. Then based on the proposed policy, a distributed SIP overload control mechanism is proposed. Besides, a probe-based SIP overload control mechanism is investigated, which achieves better performance; as for the service layer, a token-bucket based notification traffic control mechanism for IMS Presence service is proposed. Simulation and performance analysis are carried out to evaluate the proposed mechanisms.The principal contributions of the work presented in this dissertation are:(1) A sender-based end-to-end SIP overload control policy is proposed. In the previous end-to-end SIP overload control policy, the core servers of the SIP network need complex cooperation among them in order to propagate the overload information to the edge servers, which take charge of controlling overload for the SIP network. On the other hand, in the sender-based end-to-end SIP overload control policy, the core servers of the SIP network only implement local overload control that rejects requests by using503responses. Based on the received503responses, the edge servers calculate and then follow the restrictions on the traffic admitted to the network. The proposed policy can control overload efficiently as it throttles traffic at the edge of the network, minimizing the resources wasted on processing a request that will finally be rejected. Besides, the proposed policy deploys the overload control at the edge servers, without needing complex cooperation among core servers, thus it is easy to implement.(2) Based on the contribution (1), a distributed end-to-end overload control mechanism (DEOC), which controls overload for networks of SIP servers, is proposed. As an implementation of the sender-based end-to-end SIP overload control policy, the DEOC is deployed at the edge servers and controls overload based on the503responses received from the network. The paper takes the AIMD (additive increase and multiplicative decrease) congestion control algorithm of TCP as the basis in order to design the call admission rate control algorithm in DEOC. Based on the requirements of SIP overload control including aggressiveness, responsiveness, throughput and fairness, a non-linear call admission rate control algorithm is proposed. Theoretic analysis and extensive simulations verify that DEOC can keep high throughput even when the offered load exceeds the capacity of the network. Besides, it responds quickly to the sudden variations of the offered load and achieves good fairness.(3) Based on the contribution (2), a probe-based end-to-end overload control mechanism (PEOC) is proposed. Similar to the DEOC, the PEOC is a sender-based end-to-end SIP overload control and controls overload based on the503responses received from the network. Besides, by probing the SIP network with SIP messages, PEOC estimates the network load and controls the traffic admitted to the network based on the estimated load. Thus PEOC can infer the overload of the network more accurately and in a timely manner and thus can control overload more efficiently.(4) A token-bucket based notification traffic control mechanism (TNTC) is proposed, which controls the notification traffic for IMS Presence service. The TNTC aims at upgrading valid access probability while controlling the notification traffic. A mathematical model of a queuing system is proposed to describe TNTC. Based on the proposed mathematical model, the main probability features of TNTC are analyzed and the effects of different parameters on the performance of TNTC are investigated. Extensive simulations verify that TNTC can effectively control notification traffic and perform better than the existing schemes in terms of valid access probability and update arrival rate.