Cognitive and Automated Network Operations for Present and Beyond

Celtic-Next: Project CANOPY


Due to increased competition and pressure on costs, Communications Services Providers (CSPs) face significant technological evolution challenges. New strategies will be required to find the best trade-off between Quality of Experience (QoE) delivered to the end customers whilst keeping network operation expenses low. Moving forward, CSP’s should transform their current operative model aiming at more automated and self-managed processes, taking advantage of Artificial Intelligence (AI) and Machine Learning (ML) techniques to handle larger volume of requests, using less resources and relying on a team of specialized engineers and data scientists to manage the new framework.

This transformation process implies the use of Cognitive Intelligence to evolve from a reactive oriented approach towards a proactive and preventive new mode of NOC operation. By using AI/ML, it is possible to make faster decisions, processing network information in near real-time and automate network functions. These techniques rely on the detection of known patterns based on past data, in order to predict the future occurrence of similar network issues before they unveil, thus allowing to set up the required set of self-healing actions to resolve these problems, even before the customer has ever been impacted. Artificial Intelligence will play a crucial role in helping NOC teams to operate more efficiently by using specific knowledge modules that will provide extra information and guidance about anomalous information, ticket resolution and means to anticipate problems.


The objective of this project is to create a novel NOC proactive management solution that will enable the evolution from the current reactive mode of operation towards a proactive and preventive mode.

The vision is to predict problems that are going to occur before they impact customer service, providing an integrated view of the issue being solved, performing Root Cause Analysis (RCA) to understand and identify what causes triggered the problem and the corresponding recommended resolution. A smart auto-ticketing management system will identify contextual performance degradation and will automatically trigger network regenerative actions after a failure, as for instance opening trouble tickets with the right parameters. Moreover, engineers will have access to dashboards with a comprehensive set of performance indicators providing a full overview of the issues and actions taken.

Major Achievements

In summary, this project aims to achieve the following innovations: notify the engineers of problems that are going to occur before they impact customer service; provide an integrated full view of the problem being solved, with details of the alarms or predictive insights, performance measures, related historical trouble tickets, recent change requests on the component; provide the recommended resolution with step-by-step instructions to solve the problem; integrate with an automation tool to automate the resolution; improve the maturity of the service provider’s NOCs and evolution to a service operation centre (SOC) focusing them on critical customer service issues and also empowering them to become proactive and preventive versus reactive; create an automated RCA to understand and identify what causes originated a particular problem; quickly identify the right resolution based on the various sources of structured and unstructured information the algorithms are trained on; support an automated resolution of problems where applicable; integrate a smart auto-ticketing management to identify contextual performance degradation; have a network capable of regenerating itself after a failure.

Also, the project should consider a set of self-optimization algorithms to improve the network performance. An innovative solution capable to managing a proactive and preventive NOC will be developed under the project. It will analyse the context of the network and automatically open trouble tickets with the right set of parameters to perform regenerative actions.

Finally, the engineers will have a set of visualization dashboards where they can obtain a full view of the problem and the proposed solutions to solve it.