Signaling Architectures for the Interaction of the Session Initiation Protocol and Quality of Service for Internet Multimedia Applications
Goulart, Ana Elisa Pereira
MetadataShow full item record
Interactive multimedia sessions combine requirements of traditional telephony services and Internet applications. This requires call setup, call signaling, negotiation, routing, security, and network resources. Seeking to facilitate the use of quality of service (QoS) mechanisms to users of such applications, this thesis presented new signaling architectures that addressed the interaction of the Session Initiation Protocol (SIP) as the session control signaling protocol and current resource management frameworks. The Differentiated Services (DiffServ) architecture is used as the primary example. The new architectures addressed the roles of SIP agents and proxy servers in subjects such as resource negotiation, call authorization, and end-to-end QoS in heterogeneous networks. First, an architecture based on the use of QoS-enhanced SIP proxies and a SIP-based interface between the application and network layers was developed, implemented in a testbed, and performance enhancements demonstrated. Further studying of the Internet Engineering Task Force (IETF) proposal for the integration of SIP and resource management led to the development of a new signaling scheme, Resource management Overlapped with Answering Delay (ROAD). It explores the SIP user agent interaction with the network in a way that takes advantage of parallel user answering delays and reservation delays. An experimental evaluation of the ROAD scheme showed its call setup delay savings and reduced signaling load. Then, on the interaction of SIP and call admission control, an inter-domain call authorization model that implements the concepts of proxies as gate controllers (QoS-enhanced SIP proxies-GC), and that provides call authorization status and adds more granularity to the authorization process is proposed. This model showed to be scalable in terms of the need to add more resources to compensate for the increasing service load on the servers. Finally, an example framework that applies the new signaling architectures to achieve end-to-end QoS in heterogeneous networks is presented.