Synergizing Smart Network Operations with Digital Twins & Generative AI

Contributors: Guy Ben-Baruch, Trista Wu, Dan Hunt, and Fawad A. Qureshi

Yesterday, when I was at Heathrow Airport waiting to board my plane to Barcelona, I had problems with my 5G connectivity. I tried connecting to the WiFi and I received the 502 error that the wifi has too much traffic.

What can we do when ubiquitous connectivity is expected by consumers everywhere? In the fast-evolving landscape of Industry 4.0 and digital transformation, private 5G networks are the backbone, bringing advancements in various sectors. These networks tackle crucial challenges from manufacturing to enterprise campuses and airports, enhancing reliability, latency, security, customization, and scalability. However, efficient network operations remain a critical hurdle, particularly for enterprises needing more dedicated on-site engineers.

Introducing GenTwin

Enter GenTwin – a revolutionary Smart Network Operations solution at the intersection of Generative AI and Digital Twin technologies. Developed by Amazon Web Services (AWS) and Snowflake, GenTwin takes airports as an example, aiming to empower network operations teams with seamless management of their intricate networks.

Watch the solution in action in this video featuring Ryan C. Green, Guy Ben-Baruch, and Fawad A. Qureshi:

What is a Private 5G Network?

A private 5G network operates like a public 5G network, offering comparable functionalities but with the added capability for the owner to control access and utilize licensed or unlicensed wireless spectrum. Essentially, private 5G harnesses the advantages of conventional 5G within confined settings like manufacturing plants, ports, airports, campuses, or business parks.

What is a Digital Twin?

A digital twin is a virtual representation of an object or system, evolving throughout its lifecycle with real-time data updates. Leveraging simulation, machine learning, and reasoning, digital twins play a pivotal role in decision-making. Key applications of digital twins include enhancing quality, predicting maintenance needs, optimizing operations, planning upgrades, ensuring safety, optimizing assets, and simulating R&D/engineering scenarios. With Digital Twins, enterprises can efficiently run simulation modeling of their business processes without causing potential hazards in production.

What is Network Slicing?

Network slicing is a key concept in 5G telecommunications architecture that allows a single physical network infrastructure to be logically partitioned into multiple virtual networks, each tailored to serve specific applications, services, or users with varying requirements. Each network slice operates as an independent, isolated instance of the network with its own set of resources, performance characteristics, and management policies. 

The Role of Generative AI in Telecom

We have discussed before that the telecom industry is ripe for disruption from generative AI. Generative AI can be a productivity enhancer across many use cases in telecom network management across use cases such as:

• Optimizing network performance

• Developing new ideas and delivering insights

• Identifying and resolving network problems quickly and efficiently 

• Automating the design and deployment of new networks

• Predicting future network outages, spikes in traffic, or other problems

The Solution in Action

The Data Sources

There is no AI without data, so first, we need to identify the different data sources that can help us in building the solution:

1. Flight Schedules. Since we are building the solution with an airport as an example, we need to know the expected traffic. We will use OAG, the world's leading provider of digital flight information, intelligence, and analytics for airports, airlines, and travel tech companies.

2. Expected Foot Traffic. OAGonly provides schedules, so we need to know the flight occupancy rate. To overcome this, we can also include a real-time event data source such as PredictHQ to understand disruptive events, concerts, sporting events, etc.

3. Speed Test Data. To help with cell planning, we must identify areas where users face connectivity issues. We can use data from umlaut company to identify those trouble areas. Umlaut telecommunications benchmarking reports address speed, latency, and consumer experience across over 200 networks and 120+ countries. 

4. Telecom Network Data. Getting the data in real-time from the edge devices and different network data from other telecom sources will help us manage network operations.

5. 3D Model of the airport to create a Digital Twin

Technologies Used

1. Snowflake Marketplace. For getting flight schedules, events data, and speed test data into the solution.

2. Snowflake Cortex. It is a fully managed service that offers access to industry-leading AI models, LLMs, and vector search functionality, enabling organizations to analyze data and build AI applications quickly. 

3. Snowflake Geospatial. For processing large amounts of location data at scale within the Snowflake Data Cloud.

4. Streamlit. For building simple and easy-to-use interactive applications for the end user to interact with the LLM.

5. Amazon Bedrock. Amazon Bedrock is a fully managed service that offers a choice of high-performing foundation models (FMs) from leading AI companies

6. AWS IoT TwinMaker. AWS IoT TwinMaker makes it easier for developers to create digital twins of real-world systems such as buildings, factories, industrial equipment, and production lines.

7. AWS Sitewise. AWS IoT SiteWise is a managed service that makes it easy to collect, store, organize and monitor data from edge devices at scale to help you make better, data-driven decisions

Solution Flow

At the heart of GenTwin lies AWS IoT TwinMaker, providing a virtual representation of the private 5G network within the airport's context. This digital twin enables a visual understanding, allowing teams to monitor, analyze, and simulate network architecture and performance. Imagine a 3D view showcasing the distribution of 5G cells and WiFi access points throughout airport terminals and infrastructure.

Generative AI for Enhanced Efficiency

Complementing this is Amazon Bedrock's Generative AI, acting as a virtual assistant and orchestration layer. This AI-driven interface streamlines tasks for the network team, answering queries, summarizing network issues, recommending solutions, and even automating the creation of tickets for engineering teams. The RAG model has been implemented in Snowflake via the Cortex functionality.

Real-Time Insights

GenTwin utilizes real-time data, ingesting network telemetry and airport traffic data from Snowflake. This seamless integration allows near real-time monitoring and alerting capabilities for the private 5G network, enhancing situational awareness and observability. 

Data Monetization

Snowflake also provides footfall data on how people are moving through the airport. The airport operations team can use these insights to change the rental prices of different retail shops. In addition to monetization, this data can be used for better capacity management at the airport.

Solution Benefits

Imagine GenTwin as your airport's digital brain, allowing operations teams to foresee, simulate, and troubleshoot network scenarios quickly. Through intuitive interfaces like Grafana and Streamlit, the solution provides a user-friendly experience, demystifying technical complexities.

Event Planning Simplicity:

• Scenario: An upcoming event like the MWC Barcelonatriggers the need for increased network capacity.

• Solution: GenTwin enables simulation of traffic scenarios, testing solutions, and proposing changes efficiently. The AI chatbot assists in generating tickets and assigning tasks to network operation teams for seamless execution.

Real-Time Troubleshooting:

• Scenario: Automated alerts signal network performance degradation in specific areas.

• Solution: GenTwin's digital twin and Generative AI speed up troubleshooting, offering visual insights into affected areas. The chatbot provides recommendations, and the orchestration layer ensures swift issue resolution.

Future Possibilities with GenTwin

GenTwin's potential extends beyond its current capabilities. Integrating Generative AI and Digital Twins opens doors to innovative use cases such as network slicing creation and management in an emergency.

In essence, GenTwin simplifies network operations and paves the way for future innovations, where airports and industries can harness the power of intelligent technologies for unparalleled efficiency, effectiveness, and autonomy. It's not just about managing networks; it's about shaping the future of smart connectivity.

In this video, Guy Ben-Baruch and Fawad A. Qureshi discuss the solution in detail with Robert Belson.

For more information, read about Snowflake Telecom Data Cloud and AWS for Telecom.