Resource management of integrated services networks.

摘要

One of the fundamental objectives of a broadband intergrated services network is to offer existing as well as emerging heterogeneous services in an integrated, streamlined manner. To achieve this goal, a broadband network must be able to satisfy the Quality of Service (QoS) requirements of various types of services including video, voice, high-speed data as well as other new multimedia applications.; Providing QoS in an integrated services network imposes various resource management actions at different time-scales. At short time scales, related to packet bursts, congestion avoidance techniques such as Transmission Control Protocol (TCP) have been implemented. At connection or call level, connection admission control (CAC), to determine if a connection request can be accommodated (if QoS requirement can be met), is developed. If connection is accepted, routing algorithms are devised in order to find the optimal route to the destination through the network. On a less frequent basis, service providers solve capacity planning problem to determine the optimal network topology for a given set of demands. Here, the goal is the efficient management of network resources.; In our view, an integrated services network, let it be ATM (Asynchronous Transfer Mode) network, or MPLS (MultiProtocol Label Switching) enabled IP (Internet Protocol) network, has three major resources: Link Bandwidth, Number of Connections it can handle, and Call (Connection) Processing capacity.; First, we consider dynamic bandwidth allocation schemes and develop engineering guidelines for measurement-based admission control algorithms. In this respect, we provide answers to two important questions: “What to measure?” and “How to measure?” Sampling and measurement interval tuning knobs are studied extensively.; Second, we provide a noble approach to manage call processing and connection space resources in an ATM network in a challenging Voice Trunking Over ATM (VTOA) application. The scheme is based on delayed release of switched virtual connection (SVC), widely called as sticky SVCs.; Third, with a longer network planning horizon, we offer a formulation based on multi-commodity flow problem to efficiently utilize three network resources mentioned above. A simple analysis is also provided to determine the implications of resource optimization to routing.