In the past years, there has been an enormous growth of data communications. To accommodate for the fast growing demand for bandwidth, new techniques are needed to allocate and to make use of available resources more efficiently and effectively. One such promising new technique is network coding.
Unlike the traditional store-and-forward approach where intermediate nodes in networks can only replicate and forward data packets, the network coding approach allows the data packets to be modified before they are transmitted from the intermediate nodes. It was proved that using network coding, the required transmission bandwidth can be reduced substantially.
This following research subprojects addresses some of the challenges in the design and implementation of network coding in realistic scenarios.
One of the key uses of the Internet is for dissemination of large amounts of data to many users spanning across the globe. There are many applications falling into this category. For example,
This project brings together new ideas from coding theory and information theory and applies them to the problem of information distribution over the Internet. The research objective is to design a set of high quality solutions for large-scale content distribution. The approach taken is fundamentally different from conventional protocol design: rather than basing designs on TCP (transmission control protocol) flows (the basic "digital pipe'' of traditional designs), the proposal is to apply fountain codes to improve efficiency, performance, flexibility and resilience over existing protocols. The obtained solutions will also be compatible with the current Internet infrastructure, as the proposed solutions are end-to-end and peer-to-peer solutions that construct practical overlay network topologies to achieve the design objectives.
Grid technologies offer an unparallel advantage by sharing computational resources across a wide area network, allowing a community of resource donors to tackle computational problems much larger than any individual organization can tackle alone. It has thus become an essential research tool in "Big Sciences" such as subatomic physics.
This research project has two main goals. Firstly, it aims at exploring and developing general-purpose grid computing technologies that can serve as fundamental building blocks for grid computing. Secondly, we anticipate that several of the efforts will result in computational technologies and resources directly beneficial to particle physics. To these ends, we will establish the "Data and Computing Grid Laboratory" for studying, developing, and deploying new technologies to seamlessly integrate geographically distributed computing resources. The proposed research program are divided into three sub-projects described as follows.
Design of a new data transport technology on a grid that utilizes bandwidth optimally by using digital fountain codes and network coding.
Development of a language-based security mechanism to facilitate the development and deployment of secure grid applications.
Development of specialized grid applications, middleware and computational resources to facilitate the experiments and analyses in the GlueX project.