Today's networks are all based on the same fundamental principle: the network forwards data but the information itself is processed only at the end-nodes. The network coding paradigm breaks this assumption by enabling intermediate nodes to process and combine incoming packets into one or more outgoing packets. This idea has generated significant interest originally in the information theory and computer science communities and more recently in the networking community, when it was demonstrated that it can improve the performance of peer-to-peer and wireless mesh networks. However, network coding does not come for free: it adds complexity, increases processing and storage costs, and requires that we re-think the protocol stack. Currently, the full potential and limitations of network coding in the Internet are still poorly understood.
The goal of this research is to bridge this gap between the theory and practice. The project starts by identifying application areas that can benefit from network coding, including network security, multimedia delivery, and network inference. It designs novel network coding mechanisms to solve practical problems in each area and evaluates the cost-benefit tradeoffs. Building on this experience, the project then identifies common building blocks and studies deployment and architectural issues.
Intellectual Merit: the project designs innovative mechanisms based on network coding. It also studies the role of network coding in the future Internet architecture.
Broader Impacts: The project integrates graduate student training, interactions with industry and foreign academic institutions, and course development. Results will be disseminated through research publications and communication with industry.
Data networks are traditionally based on the same fundamental principle: the network forwards data but the information itself is processed only at the end-nodes. The network coding paradigm breaks this assumption by enabling intermediate nodes to combine incoming packets into one or more outgoing packets. However, network coding does not come for free: it adds complexity, increases processing and storage costs, and requires that we re-think the protocol stack. The goal of this project was to bridge the gap between the theory and practice of network coding, by applying and optimizing network coding techniques in practical settings. The intellectual merit of this project lies in the following research activities and resulting findings. (1) Cross-layer optimization of network coding and transport and application layers, specifically for wireless networks . (2) Design of solutions for pollution attacks in coded systems (3) Combining tomographic and network coding techniques to improve network measurement and inference. (4) Design of novel network coding techniques for the open problem of multiple unicasts. (5) Implementation of network coding and cooperative video streaming on smartphones. The products include several conference and journal publications in top venues, and software prototypes. The broader impact of the project includes training of students (5 PhD theses), development of new educational materials, and technology transfer (i.e, part of the project led to an I-CORPS and eventually to an SBIR project).
