Companies have now developed technology that uses deep learning to train itself on parameters to avoid the impact of systemic errors in optical signal transmission. This occurs while learning to estimate optical transmission signal parameters, a problem unique to optical communication systems, including symbol rate and OSNR (Optical Signal-to-Noise Ratio). The companies developed an experimental transmission system within Fujitsu Laboratories that emulates an optical network, and with nearly 10,000 data points, they verified that this technology could estimate OSNR with a measurement error of 1% and estimate the modulation format and symbol rate with a measurement error of 5%.
When problems arise in the construction or operation of an optical network, this technology now allows tasks to be completed in a matter of minutes, tasks that would take an expert several days using specialized measuring equipment. This will significantly simplify the construction, operation, and management of networks.

Development Background:
Communication traffic on the optical networks that underpin an ICT-driven society is expected to increase dramatically in the coming years, along with the number of internet-connected devices. To accommodate this volume of data, a number of new optical transmission technologies are being adopted in optical networks, one after another, and these networks are expected to become even more diverse and complex. Consequently, there is a demand for technologies that facilitate the construction, operation, and management of optical networks.

fujitsu-image-2-wnetworks
aiming to increase capacity and distance, the growing complexity of optical transmission signal types and device parameter configuration means that network construction or troubleshooting can take several days, creating significant challenges for building and managing fiber optic networks. As a result, there has been a demand for the development of technologies that can remotely monitor the status of optical networks to address these issues. However, challenges have arisen in measuring the information that network operators and administrators need without relying on dedicated measuring devices, due to the unique optical signal characteristics of newly deployed optical transmission technologies.
 
Regarding the technology developed,
Fujitsu Laboratories, Fujitsu Laboratories of America, and Fujitsu R&D Center have developed technology to measure the optical signal transmission parameters (signal-to-noise ratio, modulation format, and symbol rate (1)) necessary to build and operate a network from optically transmitted signals at remote optical receivers.
This newly developed technology trains a deep neural network by inputting the signals received by the optical receivers into the network. By using the measurement equipment's results to provide monitoring labels, this technology trains the deep neural network to recreate the measurement results produced by the equipment, allowing it to estimate the optical signal transmission parameters. Since systemic errors can arise in signal characteristics, such as laser frequency, when an optical transmission signal is received, if the received data is used for training as is, the neural network will be trained to specialize in erroneous states. This could increase measurement errors in the estimates. As a way to counteract this, the new technology virtually generates signals based on optically transmitted signals in varying states, for example, by virtually generating multiple datasets with different laser frequencies and then combining them to form the training dataset. By doing so, it is possible to reflect a variety of situations in the training results, allowing this technology to minimize measurement errors in the estimates.

Laboratories
, Fujitsu Laboratories of America, and Fujitsu R&D Center conducted a test by building a simulated transmission system that models a real optical network. The test verified that with 10,000 data points, this technology could estimate OSNR with a measurement error of 1% and the modulation format and symbol rate with a measurement error of 5%. With this technology, it is expected that tasks that previously required an expert to use specialized measuring devices for several days can now be estimated remotely in a matter of minutes.
Future Plans
: In the future, Fujitsu Laboratories, Fujitsu Laboratories of America, and Fujitsu R&D Center will conduct tests in a real network environment, with the goal of commercializing this technology in fiscal year 2019 or later. The companies will also continue research focused on the automated operation of optical networks.

More information
________________________________________
• [1] Symbol rate: The rate at which amplitude and phase information can be switched when modulating electrically transmitted data via light.
• [2] Coherent digital receiver: A receiver that can handle optical signal phase information with stability, which was a problem with previous coherent receivers, by applying digital signal processing to optical communication.