PLANET: Physical Layer Aware Networking

    Mission

    5G, cloud computing and DCI will drive a dramatic increase in the IP traffic over the next years. Operators aim at exploiting the existing infrastructures to maximize their return on CAPEX. The answers to these requests are moving-on towards the exploitation of data transport over the installed cables beyond the C-band and opening up network hardware and software to reduce user costs and maximize operator and vendor returns. So, the mission of the PLANET team by OptCom is to abstract multiband data transport by simulation and mathematical modelling in order to enable the physical layer awareness for multiband open optical networking and open network engineering, planning, management and controlling.

    People

    Team leader:

    Members:

     

     

     

    Research Activities

    • Open Optical Networks: Politecnico di Torino is an active member of the consortium Telecom Infra Project (TIP) promoted by Facebook that aims at the development of HW and SW for open and disaggregated wireless and optical networks. Prof. Curri represents PoliTo within the consortium and he is the scientific rensponsible of the PoliTo-TIP MSA. He leads the PLANET team activities within the OOPT working group of the TIP. The PLANET team collaborates with operators and vendors, such as Facebook, Microsoft, Orange, Telefonica, Telia, Cisco , Juniper Network, Infinera and many others, to develop mathematical modeling for the abstraction of the physical layer and to implement the outcomes within an open source python library named GNPy. Prof. Curri holds the role of GNpy Scientific Chair within the PSE-OOPT WG of the TIP. The GNpy lybrary aims at estimating and predicting the phyisical layer quality of transmission (QoT) with the purpose of promoting the implementation of open common API's for management and control of networks including the physical layer awareness. The developed source code is available in the TIP GitHub repositories OOPT-GNpy and OOPT-Raman.
    • Machine Learning Aided Networking
      • Optical line system controller: in collaboration with the Links Foundation and with vendor companies, the PLANET team investigates on the use of the artificial intelligence to optimize the controller for optical line systems.
      • Planning, management and failure recovery: relying on the knowledge on the physical layer abstraction, the PLANET team investigates on the use of the machine learning techniques in managing optical networks.
    • Physical Layer Aware Network Assessment: thanks to the physical layer abstraction enabled by the data transport models, the PLANET team has introduced the Statistical Network Assessment Process (SNAP) framework for a Monte Carlo in-line analyses on the impact of the physical layer options on networking metrics. Within such a framework, the PLANET team has also developed the Offline Physical Layer Assessment (OPLA) tools with the aim of the off-line evaluation of the impact of the physical layer merit on the network layer performances. These activities are funded by vendor companies.
    • Multiservice Optical Networks: the cohexistence of heterogeneous features such as modern high capacity transmission, backward compatibility with legacy equipment and distribution of reference unit of the standard time for metrologic purposes in the same optical network enables the reuse of the same infrastructure but it also leads to the cohexistence of strongly heterogeneous signals. In collaboration with Links Foundation, vendor companies and INRIM, the PLANET team investigates such non-trivial scenario to properly address the management, planning and optimization procedures. In particular, the PLANET team is part of the TIFOON (EMPIR Call 2018 - SRT-s21) project funded by the EU. The PLANET team is the only telecom group of the project targeting the use of telecom technologies to deliver time and frequency within optical data network infrastructures.
    • Multilayer Network Simulations: Prof. Curri is the scientific responsible of the MSA with Synopsys for the scientific activities supporting the development of the RSoft optical solution suite, so, the PLANET team operates within such a framework in developing simulation algorithms and research investigations based on the Synopsys software tools.
    • Multiband Optical Networks: Prof. Curri represents PoliTo within the H2020-MSCA-ETN Wideband Optical Network (WON). WON is a doctoral-level training network funded by the European Commission under Horizon2020 Marie Sklodowska-Curie ITN Action. The programme trains 14 early-stage researchers (ESRs) in the area of multiband optical networks through the collaboration of academic and industrial highly qualified institutions. Solutions identified within WON will enable to overcome a possible traffic-crunch by achieving a 10-fold increase in the usable optical bandwidth of single-mode fibres. Two PhD students will be trained at PoliTo as part of the PLANET team. The PLANET team has already investigated on the extension of data transport beyond the C-band with several invited talks at conferences and on major journals.
    • Power Electronics Innovation Center (PEIC): Prof. Curri is part of the interdipartimental PoliTo center PEIC led by Prof. Radu Bojoi of DENERG.The PEIC objective is to provide power conversion solutions for electric vehicles powertrains and chargers, more electric aircrafts, energy production and harvesting from renewables, smart transformers for electrical grids, more efficient variable speed drives. The PLANET team contributes to the PEIC by implementing an optical bus to remotize controlling in power electronic grids and to operate on power electronic as a cloud implementing the Internet of Power paradigm. In collaboration with the Links Foundation, the PLANET team has developed a prototype for the optical bus.

    Research Facilities

    • Optcomputing: an HPC server cluster based on multicore Xeon CPU, Nvidia GPUs and Linux OS
    • Software: research outcomes of the PLANET team are in general addressed to develop open source Python code relying on Github as repository. For machine learning analyses, we use the Tensorflow framework.
    • Lab: experiments by the PLANET team are done in collaboration with the Links Foundation exploiting the joint PoliTo-Links applied photonics experimental facilities.

    Teaching

    Dissemination

    International Academic Collaborations

    • University of Texas, Dallas
    • NUST, Pakistan

    Master's Thesis Topics

    • Abstraction of data transport in open optical networks
      • Physical layer modelling and simulation
      • Software development within the GNpy library of OOPT of the TIP.
    • Use of artificial intelligence in designing, planning, managing and controlling open optical network
    • Statistical assessment of open optical networks
    • Photonic bus to remotize control and to enable telemetry in clouds of power electronic systems