Design and Control of Next Generation Networks:A Measurement-Analytic Approach

Purdue Research Foundation, West Lafayette IN

Investigators

Abstract

Next generation high-speed networks are expected to support exciting new applications involv- ing multi-media traffic, such as video-conferencing, tele-medicine, video-on-demand, and web TV, in addition to countless other yet-to-be-conceived network applications. The viability and the continuing success of these networks will depend crucially upon their ability to offer high performance with regard to latency, delay variation, and bandwidths they can provide to these myriad of applications. Customers will demand high-quality multimedia services, and not be satisfied with the kind of high-latency, variable delay that is characteristic of the current Inter- net. The various performance requirements from the users and the sheer size of these networks make it imperative that we properly understand at a fundamental level how to design, engi- neer, and control these networks, and develop appropriate methodologies. It is also essential that these tools be evaluated in terms of complexity and accuracy using experimentation (i.e., testbed) and simulation. To address the challenges described above, we propose to develop an innovative approach that harnesses the power of combined measurement and analysis to create design and control tools for next generation networks. We plan to elaborate on the approach in the context of: performance evaluation (i.e., QoS estimation), and QoS-sensitivity estimation. These issues form the foundation required to solve key network design and control problems. Building upon our approach|combining analysis with measurements|we will then focus on on a variety of problems that have to be addressed in order for networks to support QoS: admission control, congestion control, QoS-based routing, and network design and dimensioning. Our proposed solutions to the above problems will be supported and refined by extensive empirical studies on our experimental platform. This platform is capable of supporting a variety of network technologies and traffic characteristics. Our research team has been at the forefront of the development of important results in traffic analysis, network design, and control, and is committed to creating the necessary synergy for addressing the key problems outlined above. Our team already has a significantly productive track record in previous collaborative efforts. We are planning to address a set of problems that are critically important for deploying next generation networks with the capability of offering high bandwidth and stringent QoS to users. Creating a future broadband network that is exible, efficient, robust, and controllable is essential to the viability of our economy and to the different communities within it. Hence, if our research is successful, it will have a significant impact on the delivery of services necessary to meet the diverse needs of education, business, and entertainment.