by TIP News


In the dynamic telecommunications landscape, the need for efficient and reliable optical networks has never been greater. As operators struggle to meet increasing bandwidth demands, they must optimize network designs and performance. Enter GNPy, the Gaussian noise in Python software developed by the Telecom Infra Project (TIP). Originally designed for optical network planning, this powerful tool can also be used to compare and evaluate optical network performance, enabling professionals to gain valuable insights and make informed decisions. By anticipating the use of a multi-vendor infrastructure, GNPy provides a vendor-independent physical model that describes the behavior of each element of the network starting from general fundamental parameters. Specifically, the Physical Simulation Environment (PSE) sub-group within the Open Optical & Packet Transport (OOPT) project group is developing this open, vendor-neutral planning and simulation environment that allows operators to validate vendor-specific plans and implementation goals. In this article, we’ll explore GNPy’s general approach to testing, discuss a compelling use-case scenario, and summarize its benefits for optical networking professionals.

General Test Approach with GNPy

The GNPy Python library contains a highly accurate and versatile physical model, which describes the propagation of the light signal through the various elements that make up an optical network, such as optical transceivers, fibers, optical amplifiers, ROADMs, attenuators.

When it comes to transmission quality in optical networks, there are many factors that influence its estimation and they come from each of the network elements listed above. The figure of merit that uniquely defines transmission quality is the generalized signal-to-noise ratio (GSNR), which weighs the optical power of the received signal with the optical noise introduced and accumulated along the entire path within the network.

When undertaking a test campaign for a given optical network, the first step is to translate the network characteristics into 3 main files that the GNPy software understands, which are the topology, equipment configuration and simulation parameters. The topology file contains the description of the graph in terms of network elements and their interconnections. The device configuration file describes the characteristics of the various models used for each type of network element, and finally it is necessary to define in the third file how precisely the physical model must estimate the transmission quality for a specific simulation.

Once the simulation is run, the result needs to be compared to field measurements of certain performance metrics, such as received power, optical signal-to-noise ratio (OSNR), Q-factor, or bit error rate (BER). It is essential that both the simulation and the measured data refer to the same (or at least similar) operating condition of the network, so that the comparison can be correctly made between the model estimation and the actual behavior of the infrastructure. Wanting to evaluate the worst case of operation for the optical network, it is represented by the condition in which all the spectrum is completely used (full spectrum condition), given that each channel is affected by the maximum number of interfering channels.

It is essential to underline that the transmission quality estimate produced by GNPy is inextricably dependent on the type and accuracy of the data described within these input files.

As far as the available data is concerned, they form part of both the description of the network and the measurements useful for testing the system. It is clear that it is necessary to establish the precision required of the model to carry out the simulation according to the granularity and the type of data, in order to have a degree of precision comparable with the simulator. Furthermore, depending on the use case investigated, it is advisable to proceed with a series of simulations which aim to evaluate the impact of the variation that the less accurate input parameters have on the performance estimate.

As a last step after the test, it is possible to explore the potential of the network by using GNPy in evaluating the impact of possible upgrades in terms of performance, such as adding more channels, placing amplification sites, or changing the fiber types. By running simulations using different configurations, users can compare the performance and cost implications, facilitating data-driven decision-making. GNPy’s ability to model and simulate a wide range of scenarios empowers operators to design robust and future-proof optical networks.

InterNexa Optical Network in Perú

To illustrate the capabilities of GNPy as a benchmark for optical networks, let’s consider a use case scenario in which a telecom operator intends to evaluate how its optical network can evolve into a multi-vendor perspective in order to contain costs at equal performance, or what are the minimum investment (CAPEX) to obtain the best result in terms of cost-efficiency compromise. To this end, the operator wishes to ensure that the proposed network design meets the expected capacity requirements and delivers the desired quality of service. Leveraging GNPy, the operator can simulate network performance in different scenarios, evaluate its capacity and identify potential bottlenecks or limitations.

This approach was performed on a DWDM optical network in Perú operated by InterNexa, which is an active participant in the OOPT-PSE working sub-group and at the forefront of open optical network planning and simulation. A schematic of the infrastructure is represented in the figure, showing the presence of ROADMs, EDFAs, Raman amplifiers and ROPAs. The peculiarity of this network is that the longest central fiber span is located in the middle of the desert, with an out of the ordinary degree of management and maintenance in terms of complexity. The verification of the functioning of the network and the analysis carried out with GNPy based on the available data suggest that it is possible to double the number of DWDM channels also for the critical stretch of fiber, from 20 to 40 channels, opening to a possible reduction in costs in terms of CAPEX.

InterNexa has a strong interest in an open, vendor-neutral planning and simulation application such as GNPy provided by TIP. The Politecnico di Torino represents the main academic institution working on physical layer modeling for a robust, scalable and versatile implementation, which has carried out this activity and supported InterNexa in the validation of their optical network.


GNPy, the Gaussian Noise software in Python developed by TIP, serves as an invaluable benchmarking tool for optical network insiders. Its general testing approach provides a holistic framework for analyzing and evaluating network performance, enabling users to make informed decisions. With GNPy, an early and advanced optical network planning tool, operators can simulate complex network scenarios, evaluate capacity requirements, and optimize network designs while maintaining desired quality standards.

As a ultra-long haul (ULH) network operator, it is quite important to perform pre-validations before approaching optical layer vendors and verify the minimum and expected capacity that can be supported in the optical system. Using GNPy as open optical network simulator, it is possible to envision the necessary optimizations in the architecture and topology before purchasing the network equipment.

Through a compelling use case scenario, we highlighted GNPy’s capabilities in assisting operators with new network implementations. The improvement suggested by GNPy regards the increment of the number of channels from 20 to 40, opening to a possible reduction in costs in terms of CAPEX. Leveraging GNPy’s modeling and simulation capabilities, operators can identify potential bottlenecks, improve network architecture, and compare performance under different scenarios. This allows them to make data-driven decisions, ensuring efficient and cost-effective network designs, ultimately delivering reliable, high-performance optical networks to meet the growing demands of the digital age. However, upgrades should always be supported by an assessment of the infrastructure vendors or guarantors in terms of the feasibility of the upgrade.

Contributors’ Details

  • InterNexa. An ISA company with over 20 years of experience in the provision of information technology and telecommunications services, specialized in digital solutions, with operations in Colombia, Brazil, Peru, Chile and Argentina and have a commercial presence in the United States. InterNexa delivers comprehensive solutions for connectivity, cloud, datacenter, security and managed services; focused on maximizing efficiency and productivity, and accompanying the technological evolution of your company, thus contributing to business continuity and the construction of a digitally human world.
  • Team PLANET, Department of Electronics and Telecommunications, Politecnico di Torino. Physical Layer Aware NETworking (PLANET) originates in 2015 from the larger optical communication (OptCom) group at Politecnico di Torino, Italy. It is a research team composed by researchers and PhD candidates and led by Prof. Vittorio Curri. The main areas of interest are related to the fiber optical propagation modelling and the definition of architectures in order to implement flexibility, automation and optimization on optical transmission systems. The PLANET team mission is to abstract the data transport by simulation and mathematical modelling in order to enable the physical layer awareness for open optical networking and engineering, planning, management and controlling.
  • Physical Simulation Environment (PSE). Within the larger Open Optical & Packet Transport (OOPT) project sub-group by Telecom Infra Project (TIP), this group represents the first industry-wide effort to develop an open-source multi-vendor tool for optical network planning, the Gaussian Noise software in Python (GNPy). The tool is in active development and with it, operators will no longer have to depend on their suppliers to plan routes and network capacity, but will have an independent way to lay out their requirements and simulate network conditions. Project group members — including Cisco, Meta, Juniper, Microsoft, Orange, Politecnico di Torino and Telia Company — have made great contributions to build this tool.