CAREER: Multicast Protocols and Topology Models for the Internet

点击翻译按钮获取中文摘要

The proposed research focuses on practical solutions that will advance the current Internet technology to meet the challenge of the Next Generation Internet. The educational component of the project aims to train the network experts that the industry needs. One of the main problems of the current Internet infrastructure is its inability to provide services at high Quality-of-Service (QoS) levels. The lack of such quality guarantees affects the reliability and the commercial value of the services. New protocols and architectures are needed to support the emerging applications. An impediment in the development of such architectures is our limited ability to understand, model and simulate the Internet and its topology. The proposed work tackles these two problems. First, the project proposes to develop a multicast protocol for the Internet that will support group applications with high QoS requirements such as tele-conferencing. Second, the project proposes to develop a novel method to characterize the topology of the Internet and explore its applications. The project proposes the QoSMIC Internet routing protocol to support QoS in group (multicast) communications. The researcher's Ph.D. work introduced QoSMIC, a new QoS0sensitive multicast protocol. QoSMIC differs from the previous protocols in that it identifies multiple paths and selects the one that can provide the QoS level the applications requires. In addition, QoSMIC is resource-efficient and flexible without compromising its scalability. Preliminary simulations show that QoSMIC can support 4 times as many users with a specified QoS level compared to the current protocols. There are a lot of research issues that are needed to complete the protocol. The research proposes to develop, implement and test protocol in three steps: extensive simulations, software implementation, and testing in the researcher's network campus. The long-term goal is to incorporate QoSMIC in commercial routers through an industrial collaboration. One of the most difficult tasks in the design and the testing of a new protocol is predicting its behavior in a large-scale realistic network. Thus, the research proposes to model, re-create, and predict the growth of the Internet topology. The research presents a novel way to describe topological properties of large networks using power-laws. Despite the randomness of the Internet, the researcher has discovered that some surprisingly simple power-laws can describe the Internet topology based on measurement data during the year 1998. The power-laws fit the real data very well, with correlation coefficients of 96% of higher. The rsearcher's initial set of power-laws is only the beginning of a new way to view and model the Internet. The project proposes to continue work focusing on three major research directions. First, it will identify more power-law properties and thus characterize and classify graphs effectively. Second, it will develop a new graph model for realistic topologies. Finally, it will measure and predict the growth patterns of Internet. In all three areas, software tools will be developed to analyze, generate, and predict graphs and their topological properties. The project proposes to develop a strong educational program as a vital part of the career development plan. The goal of the educational plan is to provide students with a soled theoretical background and practical experience on cutting-edge technologies. The project proposes an educational framework that is based on a) a strong theoretical and technical curriculum, b) early student involvement in research and teaching, and c) collaboration with the industry through internships and sponsorships.