ITTC Project

Complexity, Implementation, and Management Trade-offs for Traffic Aggregation in Future Networks

Project Award Date: 06-30-2000


The quality of service (QoS) experienced by the customer is determined by how well the network can allocate the resources required for optimal performance of each customer's task. A satisfactory solution for provider networks requires a set of technologies ensuring that the network control plane accurately matches the link bandwidth and buffer resources made available to the needs of the customers task.

The IP control plane is responsible for allocating network resources, monitoring their use, and ensuring that the actual resource use of a task conforms to its allocation. The thrust of this research is the implementation, evaluation, and performance comparison of several prototype IP control plane architectures for supporting specific levels of QoS.

Four prototype QoS architectures will be considered. The first is the existing RSVP approach that reserves resources for each individual customer connection across all regions of the network and is hereafter referred to as "standard microflow" RSVP. This approach can provide excellent QoS to customer tasks but also incurs significant overhead of several kinds. As a result, it is not generally considered to scale well enough for large networks.

The three other prototype architectures are each experimental hybrid RSVP and DiffServ networks that combine the QoS advantages of RSVP with the more robust scaling properties of DiffServ in various ways. The first is a straightforward combination of RSVP in the outer regions of the network and DiffServ in the inner region that are supported by the network provider. This is also known as the "Microsoft Model." The second experimental architecture contrasts with the Microsoft model by using DiffServ at the edges of the network, but RSVP to tunnel between the core nodes of the network. Thus, it is called the "tunneled aggregated RSVP model." The attraction of this approach is the resources allocated to the much less numerous aggregates of traffic rather than the immense number of individual microflows. This can potentially avoid much of the overhead and thus perhaps the scaling problems of the standard microflow RSVP model. The third experimental architecture uses RSVP in the edge of the network and DiffServ in the core, as does the Microsoft model. However, this model enables the DiffServ region of the network to process the control place RSVP messages for each microflow, while applying the DiffServ service class model at the data plane.

The result of this work will be prototype implementations of these four QoS architectures. Through the construction, evaluation, and experimental use of these prototypes, we will learn the following: the processing and scaling attributes of each architecture, what enhancements or changes in the existing protocols are required, the management complexity of each architecture, and network performance tradeoffs of each approach.


Faculty Investigator(s): Victor Frost (PI), Joseph Evans, Jerry James, Douglas Niehaus

Student Investigator(s): Vijey Jenkal, Amit Kucheria, Karthikeyan Nathillvar

Project Sponsors

Primary Sponsor(s): Sprint

Partner with ITTC

The Information and Telecommunication Technology Center at the University of Kansas has developed several assistance policies that enhance interactions between the Center and local, Kansas, or national companies. 

ITTC assistance includes initial free consulting (normally one to five hours). If additional support is needed, ITTC will offer one of the following approaches: 

Sponsored Research Agreement

Individuals and organizations can enter into agreements with KUCR/ITTC and provide funds for sponsored research to be performed at ITTC with the assistance of faculty, staff and students.

Licensing and Royalty/Equity Agreement

An ITTC goal is the development of investment-grade technologies for transfer to, and marketing by, local, Kansas, and national businesses. To enhance this process, the Center has developed flexible policies that allow for licensing, royalty, and equity arrangements to meet both the needs of ITTC and the company.

Commercialization Development

Companies with a technology need that can be satisfied with ITTC's resources can look to us for assistance. We can develop a relationship with interested partners that will provide for the development of a technology suited for commercialization.

ITTC Resource Access

ITTC resources, including computers and software systems, may be made available to Kansas companies in accordance with the Center's mission and applicable Regents and University policies.

ITTC Calendar
There are no upcoming events at this time.