DIWINE is an international research project partly funded by the European Union under its FP7 ICT Objective 1.1 – The Network of the Future. It started in January 2013 and will last until December 2015.


DIWINE considers wireless communication in a dense relay / node scenario where WNC (Wireless Network Coding) messages are flooded via dense massively air-interacting nodes in the self-contained cloud while the PHY air-interface between the terminals (sources / destinations) and the cloud is simple and uniform. A complex infrastructure cloud creates an equivalent air-interface to the terminal, which is as simple as possible. Source and destination air-interfaces are completely blind to the cloud network structure. The cloud has its own self-contained organising and processing capability.

This concept facilitates energy-efficient, high-throughput and low-latency network communication performed directly at the PHY layer, which is capable of operating in complicated, dense, randomly defined network topologies and domains. The applications of the DIWINE paradigm are generic, being relevant to complex systems ranging from intelligent transport systems to healthcare and even machine-type communication in wireless networks.

However, in order to exhibit practical, highly focused and high impact results, DIWINE concentrates on two core application / demonstration cases: (i) smart metering networks and (ii) critical industrial monitoring and control applications. To this end, DIWINE algorithms and theoretical technology will be integrated into two industrial proof-of-concept demonstration platforms targeting the aforementioned applications. Both of these applications require low-latency, dense networking solutions and are sure to be integral to future European policy and society as evidenced by recent European Commission initiatives such as EUROPE 2020.


  • All PHY complexity is handled by the cloud while keeping the source/destination PHY air-interface simple and uniform. The wireless cloud provides a “complex” PHY “service” to the simple terminals.
  • The cloud works as an independent self-contained entity that does not reveal its internal structure or PHY to the terminals; only the cloud-source / destination air-interface is revealed.
  • Information among terminals is flooded through the cloud without distinguishing (or separately routing) individual data, exploiting the paradigm of WNC / NC inside the cloud.
  • Cloud relays / nodes are dense (with massively parallel signal interactions), and may have possibly imperfect and unreliable knowledge of the interacting signal structure and topology of interacting neighbours. Current WNC techniques require full knowledge of network structure and work with a relatively small number of well-defined interacting signals.
  • Distributed PHY / WNC relaying and self-adaptation is performed internally by cloud.
  • Several recent advances in the physical layer of multi-hop networks are exploited in the cloud; concatenated interference or Z-channels including the PHY / WNC concepts: interference alignment, interference forwarding and interference neutralisation. Project lives from the paradigm change in handling interference. Interference is not always bad.
  • Processing and coding within the cloud are shaped to comply with the latest advances of the industry in term of massive node deployments for delay-sensitive applications.

Technical / research challenges

  • Design of WNC coding and processing. The cloud uses WNC which performs all medium access tasks traditionally performed by routing and MAC directly at the PHY layer with potential large performance gains. In this way the functionality traditionally provided by cross-layer design is now implemented only on the PHY.
  • WNC coding and processing for massively parallel signal interactions. The cloud contains a large node density with a large number of interacting wireless nodes creating mutual interference between every pair of nodes.
  • Distributed intelligence of the PHY cloud – design of WNC codes and processing for imperfect knowledge of signal and topology of interfering wireless nodes. Terminals do not use signalling or a priori knowledge of network structure. Only cloud nodes have internal imperfect knowledge of signals, complementary side information and network topology. Throughput / robustness / delay / memory / energy trade-off in delay-tolerant, information storage ad hoc networks.
  • Trade-off between cloud self-organisation (by signalling) vs. diversity directly at PHY / WNC level, e.g. by properly forming hierarchical decision maps at relays to provide multiple paths for the information. Distributed cloud PHY / WNC signal processing.
  • Joint optimisation of energy and throughput. Energy consumption has two aspects: energy consumption (i) of terminals when communicating with the cloud, and (ii) within the cloud.
  • Design of simple and transparent terminal interfaces to the cloud.
Copyright © 2014
Home  |  Legal notice  |  Top
Tue, 15 Nov 2016 15:06