Quality of Service provisioning for IP/MPLS networks and EPON networks

• Main Author: Yin, Yong Ning
• Other Authors: Poo Gee Swee
• Format: Theses and Dissertations
• Language: English
• Published: 2009
• Subjects: DRNTU::Engineering::Computer science and engineering::Computer systems organization::Computer-communication networks
• Online Access: https://hdl.handle.net/10356/15167
• Institution: Nanyang Technological University

Scheduling disciplines are used in network routers/switches to provide a wide range of Quality of Service (QoS) assurances. With the development of the Internet technologies, two new branches of scheduling family arise in order to provide QoS in the context of emerging networking environments. In particular, aggregate scheduling is developed to provide scalable QoS in aggregation-based high-speed networks, while dynamic bandwidth allocation (DBA) is developed to offer efficient QoS support in Ethernet passive optical network (EPON)-based access networks. This thesis contributes to the design, analysis, and evaluation of various aggregate scheduling disciplines and DBA algorithms. The first part of this work studies aggregate scheduling using the concept of route interference. A general service scheme is proposed which features wide-range implementations. We consider a general network model and define a source rate condition in terms of the route interference. We show that there exist deterministic bounds on end-to-end delay and buffer size at all nodes provided that each flow obeys its source rate condition when entering the network. Experiments show good agreement between analytical and simulation results. To exploit the advantage of fair aggregation in aggregate scheduling, we then propose a guaranteed rate fair aggregator (GRFA) for flow aggregation and show analytically the advantage of aggregate scheduling with GRFA in delivering deterministic end-to-end delay bound. Subsequently, we devise a flow aggregation capable of guaranteed rate (FA-GR) scheduling and a priority-shaper for aggregate scheduling. This approach obtains the same level of end-to-end delay bound as GRFA-based approach while achieving lower complexity and implementation cost.