Comlab's research topics cover a broad range of topics in the field of digital communications, digital signal processing, communications networks and machine learning. In particular, our current efforts are focused on three main research fields:

  • Machine learning for optical transmission and networking

Optical networks are extremely complex systems, posing problems that are difficult to solve using conventional techniques. Recently, machine learning (ML) techniques have been successfully applied to a large number of problems in optical communications. These problems include failure management, optical performance monitoring (OPM), quality of transmission (QoT) estimation, traffic prediction, among others. At Comlab, we partner with industry and other academic institutions to solve complex problems in optical communications and networking using ML techniques. Collaborators/supporters: Fapesp, Padtec, CNPq.

  • Avanced modulation and signal processing for SDM systems with coupled channels

Space-Division Multiplexing (SDM) has been accepted as the only transmission technique able to cope with the exponential traffic growth experienced in several segments of data networks, especially in inter- and intra-data-center interconnects. Besides parallel transmission of signals in bundles of single-mode fibers (SMFs), multicore fibers (MCFs) and multimode fibers (MMFs) are important candidate technologies for the SDM transmission media. In this research field we develop a suite of signal processing techniques tailored for SDM systems with coupled channels. Collaborators/supporters: Fapesp, Stanford University (USA), Aston University (UK), Eindhoven University of Technology (TUe), CNPq.

  • Networking technologies for future optical networks with space-division multiplexing

For almost three decades, wavelength-division multiplexing (WDM) has been the key technology for supporting the exponentially increasing data traffic in core optical networks. To date, network upgrades have been mainly based on activation of additional wavelength channels within a single fiber. However, the capacity of single-mode fibers (SMFs) is limited. As the routed throughput exceeds the capacity of a single SMF provided by the entire spectrum, the deployment of space-division multiplexing (SDM) - either in multiple SMFs, multi-core fibers (MCFs), or multi-mode fibers (MMFs) - is inevitable. This research line studies viable switching architectures for smoothly evolving the network towards space-division multiplexing in terrestrial and submarine networks. Collaborators/supporters: Stanford University (USA), FAPESP, CNPq.

A simulation tool for space-division multiplexing elastic optical networks

About Space-D

This software is a custom discrete event-driven simulator for Space-division Multiplexed Elastic Optical Networks (SDM-EON) with support to uncoupled channels that are represented by single-mode multi-core fibers as the links of the network.

The dynamics of the system is represented by the event chaining. The events are associated with the snapshot system actions that indicate state transitions and can trigger other events that will be executed in the future.

The events are connection requests to connect source-destination pairs of nodes in the network. The network is managed by the Control Plane, which checks if there are enough resources to accomodate all requirements before to install each connection request in the network.

This project was supported by CNPq and by FAPESP grant 2015/04382-0.