TRANSNET researchers test new method to enable intelligent optical networks

In a paper published today in Scientific Reports, TRANSNET researchers demonstrate the successful abstraction of an installed optical network, a technique that has the potential to transform how networks are intelligently controlled and managed in the future.

Optical networks form the backbone of our digital communications infrastructure, carrying data and connecting people and places around the world. Any communication over the Internet uses optical fibre technology and demand for data is growing all the time. However, the optical devices and fibres in the network have unknown exact characteristics at any instant in time. To deal with this, operators leave lots of spare capacity to make sure there are enough margins for uncertainty, but this means that the overall capacity of the optical network is reduced. What if we could measure and abstract the physical infrastructure to allow intelligent optical networks to deliver capacity where and when it is needed?

TRANSNET researchers have now developed an innovative approach to enable more dynamic control of optical networks. They apply the concept of abstraction from computer science to simplify the description of optical transmission whilst taking into account the key physical properties of the network. The abstracted model provides a more basic interpretation of how the network performs, transforming how optical networks could be managed in the future. 

To test their idea, the team conducted a series of network transmission experiments over the UK National Dark Fibre Facility (NDFF), an EPSRC funded National Research Facility that enables researchers to develop communications technologies that will support the future Internet. The NDFF comprises a dedicated fibre infrastructure linking four universities – Bristol, Cambridge, Southampton, and UCL. For this piece of work, experiments were carried out over an installed network between three of the four NDFF partners: The University of Cambridge, the University of Bristol, and University College London. Researchers at all three locations were able to consistently abstract the physical components of the network. They then demonstrated that the performance can be accurately predicted based on the abstracted experimental data, and the accuracy of the predicted performance was comparable to more computationally demanding methods.

The research shows that an abstracted network provides a simple but robust approach to measure network performance that can be easily incorporated into the management of optical networks to enable information to be transmitted over the network more effectively. The development and application of techniques like abstraction are essential if future optical networks are to deliver the fast and flexible capacity required to meet the ever-increasing data demands of a global, more digitally-dependent society.  

The work was led by TRANSNET Investigators David Ives and Seb Savory from the University of Cambridge, along with TRANSNET members from UCL (led by TRANSNET PI Polina Bayvel), and in collaboration with colleagues from the University of Bristol and KDDI Research (Japan), as part of EPSRC programmes TRANSNET and UNLOC, and the EPSRC project INSIGHT 'Introducing Insight into the Abstraction of Optical Network Infrastructures' and is published today in Scientific Reports.


Distributed abstraction and verification of an installed optical fibre network
D. J. Ives, S. Yan, L. Galdino, R. Wang, D. J. Elson, Y. Wakayama, F. J. Vaquero-Caballero, G. Saavedra, D. Lavery, R. Nejabati, P. Bayvel, D. Simeonidou and S. J. Savory
doi.org/10.1038/s41598-021-89976-w

For a full list of published TRANSNET research, including journal and conference papers, please take a look at our publications page.