DS07 - Société de l'information et de la communication

Optimal Routing and Spectrum Assignment in Realistic Flexgrid Elastic Optical Networks – FLEXOPTIM

Submission summary

A WDM flexible grid for Spectrally Flexible Optical Networks (SFONs) was standardized in 2012. FLEXOPTIM aims to develop efficient Routing and Spectrum Assignment (RSA) algorithms able to optimize in a tractable way the WDM optical spectrum use in SFONs, with arbitrary topologies and large sizes; e.g. several tens of nodes, and several hundreds of connections. RSA addresses two use cases:
- A full set of connection requests is known in advance. 0ff-line calculation takes into account the entire set. Connections are then configured from scratch. Such network reset can be carried out periodically to make the best use of transmission resources.
- Connection requests arrive on the fly. Calculations are made on-line taking into account existing traffic and connections are configured immediately after. This corresponds to the incremental “local” optimization of the network physical resources.
Algorithms have to be compatible with SDN paradigm. Hence, FLEXOPTIM will regularly interact with Orange Labs teams involved in SDN forums and standardization bodies.
The key challenge is algorithm scalability. The RSA problem is NP-Hard, much harder than the Routing and Wavelength Assignment problem for fix grid WDM networks. FLEXOPTIM shall explore new mathematical approaches reducing the number of variables to overcome the drawbacks of current methods.
FLEXOPTIM involves two teams, from the LIMOS laboratory at University Clermont-Auvergne (including the project coordinator) and IMT Atlantique (Lab-STICC and IRISA), with an in-depth expertise in respectively applied mathematics to optimization and optical network architecture.
The project includes a management work package and two technical ones. WP0 is in charge of reporting to ANR and project coordination. It also ensures exploiting and disseminating the results as well as maintaining frequent contacts with external industrial partners and optimization experts thanks to an advisory board. WP1 shall develop new optimization tools for RSA problem. WP2 shall evaluate the developed algorithms and apply them to use cases defined in close relationship with the advisory board.
WP2 shall first define Key Performance Indicators. In the first year, WP1 shall introduce new formulations for the off-line RSA problem and deliver a basic version to WP2. During a six month evaluation process, WP1 shall enhance its formulations and use WP2’s feedback to reach a stable version to be evaluated on a dedicated Orange Labs platform. By the beginning of the third year, WP1 shall provide final specifications of the off-line algorithms and WP2 assess their performance.
WP1 shall also consider novel heuristics incorporating the previous solution structure analysis and insights from WP2 for the on-line problem. First versions of on-line algorithms will be evaluated by WP2. Stable versions, adapting SDN concepts to SFON functional architecture, will be delivered for evaluation on the Orange Labs testbed and final versions will be fully specified and evaluated at the end of the project.
As a PRC project, FLEXOPTIM primarily aims at impacting research and teaching in the areas of expertise of its partners. In particular, its results shall be published in relevant journals and conferences. FLEXOPTIM intends to cross-fertilize both optimization and optical networking fields. In particular, several optimization methods studied in FLEXOPTIM could find other applications, such as dimensioning of other network types or other discrete resource allocation problems.
FLEXOPTIM should also have a significant industrial impact. Developed codes and generic data benchmarks for simulations will be made freely available. Contributions to the Open-ROADM Multi-Source Agreement should be a very suitable tool for disseminating the results regarding node control in a SFON. A final workshop will bring together scientists from both academia and industry to present the project achievements and debate on methods and open questions in SFONs.

Project coordinator

Monsieur Hervé KERIVIN (Laboratoire d'Informatique, de Modélisation et d'optimisation des Systèmes)

The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.

Partner

LIMOS U-Clermont Auvergne Laboratoire d'Informatique, de Modélisation et d'optimisation des Systèmes
IMT Atlantique Ecole Nationale Supérieure Mines Telecom Atlantique Bretagne Pays de la Loire

Help of the ANR 298,188 euros
Beginning and duration of the scientific project: December 2017 - 36 Months

Useful links