Personal tools

CodeStream

From IEETA

Jump to: navigation, search


Title CodeStream - Real-time Network Coding For Real-time Streaming
Reference PTDC/EEI-TEL/3006/2012
PI Luis Almeida
Participants José M. N. Vieira, João Barros, Ricardo Marau, André Zúquete, Simão Reis, Rafael Figueiredo, Cláudio Patrício
Funded by FCT
Global funding (€) 145,284€200,491.92 USD
RU funding (€) 40,560€55,972.8 USD
Starts 2013/05/01
Ends 2015/04/30

The dramatic increase in the consumption of multimedia content over wireless networks, such as YouTube videos over tablets or smart phones, poses enormous challenges to telecom operators worldwide. The user’s demand for greater and greater quality of experience, particularly with respect to high-definition video, can only be met by the advent of innovative solutions that can leverage both the robustness of advanced coding techniques and the timely delivery of real-time systems and their scheduling algorithms. So far, advanced coding and real-time scheduling have lived mostly in separate worlds. Taken individually, both classes of enabling technologies rely on a rich theoretical foundation, however not much is known about their interplay and the potential benefits of combined code design and real-time system optimization. This project therefore aims to bridge the gap between advanced coding and real-time system architectures, while providing fundamental knowledge and feasible solutions for multimedia content delivery over unreliable wireless networks Focusing on multicast scenarios with multiple concurrent transmitters and receivers, we shall investigate how network coding, i.e. the ability to mix different packets and information flows by means of algebraic operations, can be used to increase the reliability of packet delivery under severe time constraints and strict deadlines for data reception. The main challenge to be overcome is the end- to-end delay incurred by coding schemes, which are typically optimized for throughput and therefore insensitive to the time it takes for the receiver to have enough packets in his buffer to be able to decode. The key idea that can solve this problem is the intelligent use of feedback and packet acknowledgements.