The Internet does not currently provide end-to-end Quality of Service (QoS) guarantees across multiple network providers. We demonstrate that networks. by using dynamic pricing, can provide end-to-end QoS guarantees for those applications that need it.
We propose the concept of QoS-Transit Services: a set of primitive services offered by an Internet Service Provider (ISP) in order to deliver packets with statistical performance guarantees within its network, using dynamic pricing. ISPs can choose their own pricing schemes, as long as QoS is guaranteed. Through simulation, we demonstrate that even simple pricing mechanisms can guarantee the advertised performance of a QoS-Transit Service. End-to-end QoS across multiple networks is achieved by using multiple QoS¬Transit Services, from different ISPs, in sequence.
Since QoS-Transit Services are offered by ISPs to generate revenues. we determine how to allocate bandwidth among the services in order to maximize revenue. assuming that demand functions can be estimated. We propose the IterLP and IterGreedy heuristics to determine the optimal allocation of bandwidth on predefined paths. IterLP achieves
revenue close to 99% of the optimal solution. achieving this result quickly. IterGreedy achieves 90-95% optimality, but executes faster than IterLP.
Additionally, to determine the paths on which to route the QoS-Transit Services so as to maximize the ISP's revenue, we propose three heuristics with different specific advantages: Service Grouping. Iterative Bottleneck Avoidance, and Iterative Bottleneck Avoidance with Tabu. We demonstrate that Iterative Bottleneck Avoidance with Tabu achieves approximately 98% of an optimal solution. Route selection is also shown to be more important when fewer QoS-Transit Services are offered.
When demand functions cannot be adequately estimated. an ISP can use our Iterative Allocation Adjustment heuristic to find the optimal allocation of bandwidth for a set QoS-Transit Services. The heuristic achieves over 95% of the optimal revenue for an ISP. We then examine how better routes can be identified by valuing the links in the network to identify rerouting possibilities, or to identify paths for new QoS-Transit Services.
To receive a specified end-to-end QoS, customers may use concatenated QoS-Transit Services. We propose the Hub Charging Model to provide scalable charging, using Brokers as middle-men. Additionally, we propose the concept of Overlay ISPs: ISPs that provide QoS-Transit Services by controlling an overlay network. An Overlay ISP, a type of virtual network operator, can offer QoS over a large geographic area without universal ISP support for QoS-Transit Services.
Finally, we detail an architecture for offering QoS-Transit Services using existing protocol standards. By using a Broker-ISP as an intermediary between customers and the 1SPs offering QoS-Transit Services, routing and charging complexities can be hidden from the customers. Additionally we describe how security, billing, metering. and policing can be achieved.
The competitive environment assumed within this dissertation is now emerging, with ISPs charging content providers in order to deliver content at a superior quality level to the ISP's customers. The time is therefore ripe for dynamic pricing to be deployed.
| Identifer | oai:union.ndltd.org:uvic.ca/oai:dspace.library.uvic.ca:1828/2177 |
| Date | 09 February 2010 |
| Creators | Shelford, Steven John Roy |
| Contributors | Manning, Eric, Shoja, Gholamali C |
| Source Sets | University of Victoria |
| Language | English, English |
| Detected Language | English |
| Type | Thesis |
| Rights | Available to the World Wide Web |
Page generated in 0.002 seconds