Accelerating quality-assured 5G RAN rollout
This proof of concept project is looking at ways to accelerate the implementation of 5G RANs to to address the time-consuming drags on network deployment.
Accelerating quality-assured 5G RAN rollout
Telco operations teams have built generation after generation of networks across every terrain and in almost every culture and economy. Collectively, communications networks should be one of the great wonders of the world. However, there is no operations team on the planet that can deploy 5G networks within the timeframe or at the scale necessary meet demand without far greater reliance on automation. A TM Forum Catalyst proof of concept called Time Crunching 5G RAN Rollout with AI is looking at ways to accelerate the implementation of 5G RANs. The team demonstrated its project on July 30th as part of TM Forum’s Catalyst Digital Showcase. The project is championed by Claro. It seeks to address the time-consuming drags on network deployment stemming from the old practices of long and redundant daily meetings, unnecessary repeated site visits, little oversight, manual inventory resource management, and outdated RAN tuning methods. At the same time, the project shows how artificial intelligence (AI), machine learning and analytics can reduce costs and eliminate redundant work.
“We believe the only way to obtain differentiating results in CapEx and time savings and at the same pace expand our footprint and assure quality of increasing types of services is to transform our current working methods and challenge the status quo in our daily network rollout process using state of the art technology,” says Hugo Salazar, engineering and implementation director of Claro Colombia.
Targeting four use cases
Claro has been an active champion in the project, working with participants Blue Prism, Brightcomms, Creativity Software, everis, an NTT company, Microsoft, Nokia and VIAVI. The group addressed four targeted use cases by connecting infrastructure for each of the four proof-of-concept environments into Claro’s labs system. The team had the following objectives:
- Accelerate site survey processes by using images taken by technicians with mobile devices of hardware configurations which are stored in inventory systems.
- Leverage AI, machine learning and analytics for image interpretation of these images to accelerate site verification during the installation process, thereby reducing costs and repeated site visits
- Automate processes by implementing robotic process automation (RPA), big data and analytics to support closed-loop On-air processes
- Speed the initial cluster tuning phase using self-organizing networks (SON), single-site verification tools, machine learning, and big data analytics
“We knew our current process of quality assurance in the network and logical parts of the rollout had opportunity for improvement around digitalization,” Salazar says.
General Architecture of the Catalyst project
Addressing the challenges
To meet its goals, the team faced five main challenges described below. While each is significant on its own, Salazar notes that time-to-market and initial cluster tuning are probably the most important.
“[Solving them] will bring operational efficiencies to the operator and I am convinced they will be key to differentiation in the market in the future,” he adds.
- Site density – with the number of users and devices set to increase ten-fold over the coming years, operators must scale the number of radios and increase site density in general to support so many connections.
- Investment growth & budget execution – operators cannot match a ten-fold increase in density with a like increase in their infrastructure investment.
- Time to market – hyperscale cloud providers and over-the-top content providers are beating operators to market with innovative new services, making time to market a serious challenge.
- Competitiveness – not only are these competitors getting to market quicker, they are doing so with more flexible business models and operations. Operators need to be both fast and responsive.
- Initial RAN tuning – as a template-oriented activity, cluster tuning is a laborious process that does initial set up and testing for each individual site as well as setting up sub processes such as data collection, analysis and reporting. The scale needed in 5G deployments will put pressure on this practice.
It’s harder with 5G slicing
Hugo Nava, Regional Network Services Director at everis, says 5G slicing will further complicate the initial tuning process. “We need to carry out tuning in a more automated way,” he says. “And we need to designate in slices which are the most appropriate KPIs during the initial tuning.”
Getting the tuning right is key to optimizing the end-to-end rollout process. It helps set a baseline for performance that engineers can work to as they add additional sites. everis, an NTT company, acts as the primary integrator of the project, while Nokia provides infrastructure and software solutions (as it already does for the Claro network). Blue Prism adds robotic process automation (RPA) capabilities for emulating human activities under the Digital Workforce concept and VIAVI brings into play tools for user plane analytics techniques aiming to validate the logical implementation and ensure the user experience. They also determine the proper performance of KPIs. Microsoft Azure provides Big Data and processing tools and AI for image interpretation. Brightcomms adds engineering tools, and Creativity Software has contributed its Big Data ingestion and normalization tools. Nava says that whoever arrives first with this kind of accelerated, automated service will be a market leader. He adds that the Catalyst team will evaluate the results of the proof of concept before deciding whether to pursue a subsequent phase.
“There is still a lot to do in this area,” Nava says. “There will be space to go into a Phase 2 or even Phase 3.”
See Hugo Salazar and Hugo Nava discuss the Catalyst in the video below: