logo_header
  • Topics
  • Research & Analysis
  • Features & Opinion
  • Webinars & Podcasts
  • Videos
  • Dtw

Inception: Digital twins for 5G network infrastructure-sharing

Last year, China Telecom and China Unicom announced an agreement to share a 5G radio access network (RAN) in China. This Catalyst will address some of the key challenges relating to the planning, construction, maintenance and operation of shared 5G network infrastructure.

Annie Turner
14 Jul 2020
Inception: Digital twins for 5G network infrastructure-sharing

Inception: Digital twins for 5G network infrastructure-sharing

Last year, China Telecom and China Unicom announced an agreement to share a 5G radio access network (RAN) in China. Their intention is to speed deployment, reduce costs, and boost operational efficiencies and customer experience, particularly for enterprise customers. Communications service providers (CSPs) around the world are following suit, supporting their separate networks, operations and businesses on the same infrastructure. This Catalyst addresses how to maximize the advantages of that strategy.
The Catalyst, Inception: Digital twins for 5G network infrastructure-sharing, has China Telecom and China Unicom as champions with Asiainfo Technologies, Si-TECH, Tianyuan DIC and ZTE participating. On July 16 they demonstrated at TM Forum’s Digital Catalyst Showcase ways to deliver the greatest benefits to CSPs, their partners and customers.

This is critical because the cost of 5G base stations is about 2.5 times higher than 4G, and they use more power, but CSPs the world over are keen to leverage 5G’s new capabilities to generate new revenues. This puts them under double pressure to speed up 5G deployment and reduce costs.

5G and the power of two


The joint approach means one operator builds out a site that is used by both, which greatly speeds up construction and cuts costs. Jun Zhu, Chief Architect and Senior Product Director at AsiaInfo, explains, “We leverage the digital twin technology to simulate the world making sure every step, from network planning to operation and excellent customer experience is cost effective. We also use the twins to leverage federated learning gained from analysing data using AI, but without exposing information”.
To gain real time perception, the team deployed three digital twins to emulate the customer, network and spatial aspects (the physical environment around RAN sites) of jointly building and operating the RAN. Each digital twin works both independently and with the others via a knowledge graph on a big data platform. The big data platform is implemented as a container-based (cloud-native) distributed object, sitting at the edge of the network. ZTE implemented the network twin and AsiaInfo the customer and spatial twins.
Customer experience management needs complete visibility of data across the whole lifecycle of network, customer and spatial, and to synchronize it with what is happening in the real world. The team used TM Forum’s Customer Experience Lifecycle Metrics to collect the data and an apply emotional connection score algorithm so the customer twin could assess customer experience. The Forum’s City as a Platform Architecture and Supporting Capabilities was used to build the spatial digital twin that mimics physical entities.

AI is used to automate the network for different scenarios and the team leverages TM Forum’s AI Toolkit to analyze customers’ profiles and behavior, and improve the digital twins’ performance. The digital twins are called on to supply three sorts of data: single instance or historic; homogenous or segmented (such as to study trends in a certain age group in specific cities); and hybrid, which combines data from the network and customer teams. The different kinds of data are adopted and applied intelligently to network operation via the knowledge graph.

The common architecture allows data models to be trained without the CSPs sharing their data. Each creates a model from their data, then the two parts are merged using aggregated data and they share the federated learning.

The operations platform sits on top of this architecture layer, and is responsible for the network operation lifecycle. Federated learning can be used by both CSPs for cost effective network planning of their own infrastructure, autonomous network operations to boost operational efficiency, and personalized customer experience.

These innovations are shown through four use cases inspired by the pandemic, when the best possible use of resources is essential to meet surging demand:

  • Many countries have built temporary hospitals within days and shared infrastructure as the fastest way to deploy the necessary communications to them. The requirements can all be addressed using federated learning from the digital twins of multiple operators to simulate the scenario using immersive technologies.

  • Network optimization for video conferencing is enabled through enhanced mobile broadband (eMBB). The network twin adjusts 5G bandwidth so doctors can consult remotely via live video conferences, drawing customers’ experience of the network, scored by the customer twin, to tune network performance.

  • Network optimization for streaming data from intensive care given in ambulances is delivered via network slicing to supports ultra-reliable low-latency communication (uRLLC), where the speed of the vehicle and route information are critical, provided by the spatial twin. This shows edge-cloud collaboration for real time patient monitoring and health data collection.

  • The network twins can adjust 5G network slicing dynamically, according to the population in the hospital. As it learns the workload pattern, the twin can automatically adjust energy consumption, turning off elements when they are not expected to be in use.


Beyond these specific use cases, the digital twin-based solutions could be deployed as a general approach to smart city management.

Now available on demand:
Watch the Catalyst team showcase their project