Software and Virtualization Techniques for the Improvement of Performance and Scalability in the Integration of SDN and Network Services based on the Cloud in 5G Technologies

IMDEA Networks and Telcaria are the project coordinators
  • Financed by: Grants aimed to the execution of Industrial PhDs within the Autonomous Region of Madrid (2017). Department of Education and Innovation. Regional Government of Madrid
  • Duration: February 2018 to March 2019
  • Contact: José Felix KUKIELKA, Principal Investigator for IMDEA Networks

Today’s networks have been provisioned statically because of historical reasons, however current and future traffic trends require a dynamic way of providing processing inside the network that can span from mobiles, access networks, core networks and clouds.

In this project new incremental dynamic mechanisms are proposed in order to allow that processing be deployed whenever and wherever it is needed. The ideas presented here aims to achieve the following key characteristics that together will help enable dynamic processing in 5G networks:

  • Location-independence, so that the processing can be placed in a number of different places and networks along the end-to-end path, as deemed optimal by the beneficiary of that processing.
  • Time-independence, whereby processing can be deployed or moved near instantaneously, without end-users noticing or traffic being affected;
  • Scale independence to achieve seamless scaling by decoupling network services from their scaling.
  • Hardware independence, such that the processing can run efficiently irrespective of the different kinds of underlying, possibly heterogeneous, both commodity and proprietary hardware.

The key is to allow software basic blocks to run and migrate seamlessly in a multitude of locations in the network, with minimum disruption to traffic. However, it is possible that the network will also have a number of “static” basic blocks that cannot be moved and include switching infrastructure, Radio Access Networks, and heterogeneous commodity and proprietary hardware deployment. In such cases, it would be preferable to move higher-level programs than relying on the statically deployed functionality, thus making an efficient use of heterogeneous hardware to achieve high performance.