pjames

Tags:

• Today, the data that travels through the communication networks is like cars that travel through a highway: they enter on one end and come out of the other end (hopefully) unaltered.

3 April 2014: A decade ago, researchers realised that data communication could be much more efficient if instead of passing the data unaltered from one end to the other of the communication links, sets of the data (in the form of packets) were scrambled together in one end and unscrambled at the other end. Moreover, they counter-intuitively realized that the best way of scrambling these packets were by doing it in a completely random fashion. They named this way of routing data network coding. That result soon drew the attention of the research community and several other works followed, opening exciting directions for the use of network coding to provide significant benefits at various stages of the network design.

In the research teams at Telefonica R&D, we have been exploring the benefits of network coding at virtually all levels of the communication stack: applications (video streaming), transport (TCP), and more low level aspects such as routing (mesh networks) or even physical layer communications (wireless security).

In particular, we set out to answer the following practical questions:

• How can we increase throughput and reliability of TCP communications in low quality wireless settings?
• Having knowledge about the network topology (e.g., in the case of multiple clients willing to consume multicast video in a controlled network), how do we to assign the available resources to achieve the optimal rate to every user?
• How can we efficiently serve a common video for users with different connection qualities and access rights in a wireless scenario?
• How can we provide incentives in peer-to-peer networks where participating users have heterogeneous requirements, while ensuring quality of service?

Motivated by these observations, we developed CoMP, a network coding multipath forwarding scheme that improves reliability and performance of TCP sessions in wireless mesh networks. We also analysed the case of degraded multicasting that is, the case where different users require different subsets of the source content, and implemented a system architecture for network coding- based multi-resolution video streaming.

We also developed a peer-to-peer video streaming protocol that provides incentives for live streaming scenarios to heterogeneous users. To this end, we designed an efficient streaming system for live video over peer-to-peer networks. Such a system accommodates large populations of heterogeneous users, behave robustly irrespective of user dynamics and ensures prescribed levels of quality of experience. This is only a subset of the research performed by Telefonica I+D in the field of network coding.