Routing support for providing guaranteed end-to-end quality-of-service.

摘要

The up-coming Gbps high-speed networks are expected to support a wide range of real-time, communication-intensive applications. The quality-of-service (QoS) requirements for the timely delivery of multimedia information raise new challenges for the development of integrated-service broadband networks. One of the key issues is QoS routing, which allows selecting network routes with sufficient resources for requested QoS parameters. The goal of QoS routing solutions is two-fold: satisfying the QoS requirements for every admitted connection and achieving global efficiency in resource utilization.; Many unicast/multicast QoS routing algorithms were published recently. However, there still exist a lot of unsolved problems in this area. A few examples are listed as follows. (1) There lacks a simple solution with predictable performance and adjustable overhead for the NP-complete multi-constraint routing problem. (2) All existing algorithms are tailored towards specific problems, and there lacks a simple, general routing framework which can be easily extended to handle new problems. (3) Most routing algorithms assume the availability of precise state information about the network, which however is impractical in the real world.; We address the above problems, and the goal of this dissertation is to provide simple, general and extensible solutions for QoS routing. We study different routing strategies, compare them and outline the challenges. We propose various algorithms based on different network state models, evaluate these algorithms by analysis and simulation, discuss their strengths and weaknesses of different routing strategies, and compare them with the existing algorithms. The major achievement of this dissertation is outlined in the following. (1) A heuristic approach is proposed to solve the multi-constraint routing problem. It allows the dynamic tradeoff between performance and overhead. (2) A distributed routing framework is proposed to integrate a family of routing algorithms which support applications with QoS requirements on bandwidth, delay, delay jitter, cost, path length, and their combination. (3) Source and distributed routing algorithms axe proposed to work with state information which has a high degree of imprecision. (4) Distributed QoS routing algorithms are proposed for mobile ad-hoc networks whose topologies change as nodes move, join, or leave the networks. (5) An integrated framework is proposed to support routing and scheduling of co-existing QoS and best-effort flows.