Autonomous Network Operations

Objectives

This project focuses on enhancing network management and operations through the automation of various tasks.

Through our Autonomous Network Operations (ANO) project, we target to create AI-based software that handles and effectively supports most of the network operation tasks, extending network capabilities and simplifying the way networks are operated.

ANO also focuses on developing neural networks and algorithms that interpret the various types of data and signals generated by telecom equipment. ANO's plug-and-play AI software enables service providers to receive proactive and business-relevant insights. These insights aid in accurately diagnosing, resolving and optimizing their networks.

Research breakthroughs

Our ANO research project has achieved a number of significant breakthroughs, including the following:

The ANO project leverages both state-of-the-art AI techniques as well as our strong Nokia telecom domain expertise to deliver unique technology-accurate and business-relevant insights, allowing service providers to trustfully delegate some of the network operation tasks to software. In our own customers’ words, the gain in efficiency due to the offered level of interpretation along with the “first action right” is an enabler to reduce OPEX costs.

By creating a digital twin of a given communication network, we go beyond the limitations that appear in the physical world (limitations in data availability/existence, in scenarios, etc.) to craft cutting-edge, ML-based features enhancing the reality.

1. In our pursuit to automate the network operation, integration of network sensing capabilities is a key element. Generating insights based on network data interpretation is an important building block. We have create AI sensors that are able to deliver very accurate insights.
2. Based on our research work on training methods, data production and augmentation, our deep learning models are fully pre-trained and suitable to operate on most of the known communication technologies. This enables us to propose “plug and play” features that do not require gathering user data, online learning, adaptation down-time nor parameterization.
3. From a software perspective, our features are available through micro-services that are called via a single-point API. Regardless of use case, the technology in use or the available data (generated either by the network infrastructure equipment or by consumer devices), our software autonomously provides the most relevant insights, therefore abstracting out the complexity of networks for any end users.

Real world applications

Some applications of the ANO research are in the areas of automated network troubleshooting, network maintenance and forecasting, and intelligent network data storage:

1. Automated network troubleshooting: These developments are available in various Nokia products and platforms aiming to offering network diagnosis/analytics and predictive care. Today, our solution is live country-wide in top-tier operators in Europe and North America. It has been publicly stated by service providers that the AI-based Insights from Nokia Bell Labs have given operators the ability to proactively address issues, reducing customer calls by solving multiple issues in a single intervention and creating overall efficiencies in our troubleshooting process.
2. Network maintenance and forecasting: Migrating from one technology to another is not an easy task, and both planning and forecasting need to be done carefully. Thanks to our algorithms and accurate Insights, such tasks have been supported in real-world networks.

Future research

The future of network automation is incredibly promising, with more breakthroughs on the horizon. As the complexity of networks continues to increase and, in parallel, as the need for bandwidth and reliability increase too, the only way to proceed is to delegate tasks to accurate and intelligent software. We are committed to exploring advanced machine learning techniques, self-healing network solutions, achieving the intelligent edge for different verticals. Join us on this journey toward a more connected and efficient future.

[Axel VAN DAMME, Nicolas DUPUIS, Philippe DIERICKX] APPARATUS AND METHOD FOR GENERATING AN ESTIMATE OF A CHANNEL FREQUENCY RESPONSE. Publication number: 20230032603.

[Olivier DELABY, Nicolas DUPUIS, Axel VAN DAMME] METHOD AND APPARATUS FOR DETERMINING THE LOCATION OF IMPAIRMENTS ON A LINE OF A WIRED NETWORK. Publication number: 20230035180.

[Nicolas DUPUIS, Gert-Jan STOCKMAN, Philippe DIERICKX, Paschalis TSIAFLAKIS] METHOD AND APPARATUS FOR MONITORING A COMMUNICATION LINE. Publication number: 20210250062.

[Nicolas DUPUIS, Axel VAN DAMME] SYSTEM AND RELATED METHOD TO DIAGNOSE NOISE AFFECTING IMPAIRMENTS USING DEEP LEARNING. Publication number: 20210320689.

[Nicolas Dupuis, Axel Van Damme] Method and apparatus for monitoring a telecommunication network. Patent number: 11146344.

[Nicolas Dupuis, Axel Van Damme] Method and apparatus for determining a bridged tap length. Patent number: 10938981.

[Nicolas DUPUIS, Philippe DIERICKX] Method and apparatus for predicting the bitrate of a repaired communication channel. Patent number: 10812206.

[Dierickx, Van Damme, Delaby, Dupuis] Comparison Between CNN, ViT and CCT for Channel Frequency Response Interpretation and Application to G.Fast, IEEE Access.

[Dupuis, Rausch] POST Luxembourg utilizes AI insights to improve the home broadband experience, https://onestore.nokia.com/asset/206784

[Dupuis] How AI and deep learning can breathe new life into DSL, https://www.nokia.com/blog/how-ai-and-deep-learning-can-breathe-new-life-into-dsl/ .