AI Enables Network Operations

This post has been republished via RSS; it originally appeared at: New blog articles in Microsoft Community Hub.

Telecommunication operators run some of the most complicated and demanding networks in the world. They manage thousands of network devices from a variety of vendors while simultaneously delivering traffic to and from an ever-increasing number of endpoints, from cell phone handsets to billions of IoT endpoints connected by 5G.  All these devices combined generate data at the scale of petabytes per day in a variety of formats making it challenging to easily identify and resolve problems impacting a small number of endpoints.


Artificial Intelligence (AI) powers today’s solutions to these challenges. Azure Operator Insights is an AI-based, fully managed service that simplifies the ingestion, transformation, and analysis of data from massive and complex distributed telecom networks. It is based on a modern data mesh architecture designed for performance and cost efficiency.  Operator Insights is specifically designed for telecommunication operator-specific workloads, helping operators understand and address complex network scenarios such as the health of their networks and the quality of their subscribers’ experiences.


The Operator Insights platform provides low latency ingestion, transformation, and insights with Copilot capability.  This foundation can then be bundled up into composable units called data products by the data product factory. Further these data products can be published by operators and partners to the Azure Marketplace, and this makes up the Azure Operator Insights ecosystem.  The result is that every operator can create their own data mesh of data products, enabling the operator to correlate their data sources and derive valuable and actionable network insights.  The rest of this article reviews the data product factory and the Copilot.




Data Product Factory

The data product factory is a framework allowing partners to easily design and create new data products for the Azure Operator Insights platform offering benefits such as:

  • Acceleration of time to business value with “low code” techniques that allow rapid onboarding of new data sources from operator networks.
  • Simpler migration of on-premises analytics pipelines to Azure.  
  • Easy integration with partners’ own value-add solutions through open and consistent interfaces, such as:
    • Integration into workflow and ticketing systems empowering automation based on AI-generated insights. 
    • Integration into network-wide management solutions such as OSS/NMS platforms. 
  • Opportunity to participate in Microsoft’s growing ecosystem of data products, which will be the go-to place for operators to manage their data estate. 
  • Easy monetization of data products by selling to multiple customers in the Azure Marketplace.
  • Unique customization of Copilot knowledgebase and troubleshooting assistant capabilities.

The following screenshot demonstrates the data product factory’s approach to designing a data product. 




 The data product factory provides standardization in several key areas: 

  • Design of data pipelines for ingesting data, transforming it, and generating insights. 
  • Extensions for Copilot in Azure Operator Insights for providing context-aware knowledgebases and troubleshooting assistants using generative AI. 
  • Configuration of Microsoft Purview catalogs for data governance. 
  • Production of standardized sets of data quality metrics. 

Partners such as Amdocs and Accenture have used the data product factory to build data products to analyze network functions and customer experiences.


“Amdocs is proud to be one of the first collaborators to use Azure Operator Insights data product factory, initial targets are data products for Amdocs' 5G Core Policy and Charging functions, that we can offer through the Azure Marketplace.  Our integration of these, alongside other 5G and RAN data products, with Amdocs’ Network Platform for Operations enables the realization of operational use cases such as driving full end to end network insights for predictive maintenance and early visibility of network changes, leading to more efficient, reliable operations and an improved quality of experience for customers,” Niall Byrne, VP Network & Cloud Partner Strategy.


"Accenture collaborates with Microsoft to accelerate cloud modernization for operators. By harnessing Azure Operator Insights’ data mesh architecture, Accenture enables seamless delivery of network insights. Additionally, our Crowdsourcing and Drive Test solution, utilizing Azure Operator Insights’ Data Product Factory, provides operators an external viewpoint on their network quality. This perspective can be harmonized with other datasets, including Azure Operator Insights Mobile Core QoE data, with the ultimate aim of identifying, isolating, and optimizing the customer experiences.”, Hakan Ekmen, Global Networks Lead, Comms Industry, Accenture


The data product factory in Azure Operator Insights enables customers and partners to integrate, customize the analysis of hundreds of terabytes of network data in near real time.  And this can be gathered and correlated automatically to assist with remediation.


Copilot in Azure Operator Insights

Together with Copilot in Azure Operator Insights, AI can assess the effects of policies and procedures, disseminating network domain knowledge more widely within the network operations team, ultimately leading to improved outcomes.  Finding the “one in ten thousand of customers” that are impacted is one of an operator’s most challenging issues to solve.


Copilot in Azure Operator Insights leverages the latest advancements in AI to further enhance the network data. Unlocking the power of large language models in telecom specific operational use-cases requires more than just a traditional foundry model implementation. Copilot in AOI is building 3 main feature sets to unlock the true potential of data products and AI.




The knowledge base feature leverages the power of OpenAI GPT models with a retrieval augmented generation (RAG) approach to contextualize the responses your operators receive. The troubleshooting assistant brings in the ability for the system to be network aware. This feature is where we leverage the strengths of the data products and big data platform to go from machine learning insights into contextual alerts or notifications saving an operator considerable correlation time when attempting to mitigate on ongoing complex network issues.


The root cause assistant will further build on the Copilot in AOI's abilities by connecting past incidents and mitigation steps along with topology and suggested actions for upcoming issues.


To enable these features, we are building a powerful platform around AI technology as shown below:



These features are uniquely built for each operator and grounded in our responsible AI principles so you can leverage the power of your teams and systems to create a continually improving system allowing all your engineers to become your most proficient problem-solver, efficiently identifying, mitigating, and resolving network issues to deliver unique customer experience and satisfaction.



In conclusion, Azure Operators Insights helps operators understand and analyze an ocean of data from their network, enabling them to get timely insights.  Further, with Copilot in Azure Operator Insights, the organization can more efficiently disseminate knowledge across all employees and enable faster investigation of challenging issues.

Leave a Reply

Your email address will not be published. Required fields are marked *


This site uses Akismet to reduce spam. Learn how your comment data is processed.