Network coding for multipath TCP and adaptive redundancy
Nadia Boukhatem
Computer Science and Networks Department
February 2012
1. Context
Network coding breaks the common assumption that the intermediate network nodes have to forward the exact same data that they receive. Instead,nodes may combine several input packets of data into one or several output packets to achieve better network throughput.
Research studies showed that,in specific scenarios,network coding can provide better throughput than pure forwarding.
Network coding can be applied in many scenarios:multicast or unicast,in wired and wireless networks. It is particularly suitable for wireless networks as the links are subject to losses,interference and jamming and because of the broadcast nature of the medium.
Research on network coding has been around for a good part of the last decade with a rapid growth in the theory and potential applications.
Despite its potential,we still seem far from seeing widespread network coding implementations across networks.
In this work,we aim at applying network coding in wireless environments and contribute to bring the ideas of network coding into practice. The main objectives of this works are as follows:
2. Expected work
- First,we are interested in achieving better multipath communication throughput over lossy wireless links as well as reliability by applying network and source coding. The study will focus on multi-path TCP. The impact of path characteristics on system performance will be analyzed and predictive scheduling mechanisms for facilitating packet decoding upon arrival will be investigated. Additionally,we aim at studying how the loss rate can be optimized in case of mobility.
- Second,we are interested in investigating adaptive mechanisms for network coding redundancy taking into account the changing quality of the wireless links.
- Third,we focus on the cost of coding in terms of energy consumption. The aim is to build cooperative organisations for network coding capable mobile nodes and to distribute the coding operations over a set of nodes while optimizing the throughput and taken into account the energy consumption.
- Finally,we aim at validating the mechanisms on practical experiments.
This work will be partially achieved in collaboration with UCLA (University of California at Los Angeles).
3. How to apply?
Please contact:
Pr Nadia Boukhatem
Télécom ParisTech –Computer and Network Department
Nadia.boukhatem@telecom-paristech.fr
+33 01 45 81 82 16
