Accelerating digital transformation through automation in the 5G era

The digital economy will benefit from faster 5G deployment but to do that, communications service providers (CSPs) must accelerate their own digital transformation and innovation. This proof of concept Catalyst project explores how autonomous networks (AN) can support more intelligent and robust communications services to drive digital innovations, underpinned by artificial intelligence (AI) to realize the full potential of AN. The Catalyst project, AI for AN: Accelerating digital transformation in 5G era, is championed by China Unicom, and supported by participants Beijing University of Posts and Telecommunications, CAICT, GuoChuang Software, and Huawei. It addresses two of the most difficult issues for CSPs today: How to generate more revenue for CSPs and their ecosystem partners from the network; and making the network intelligent and more responsive to the market. Dr. Lei Lei is the technical expert at the Intelligent Network & Innovation Center in China Unicom’s headquarters and leader of the Catalyst project. He says, “China Unicom has already adopted AI for full lifecycle management of 5G networks, from network planning to construction, maintenance, and optimization. We have achieved a series of business innovations and great success in the market. This Catalyst project shows how and calls for cross-industries collaboration to further promote the use of AI for AN.”

Intelligence at scale

A good example is China Unicom’s collaboration with Tencent, owner of WeChat and games such as the Arena Of Valor. The CSP can tweak the network’s performance, in real time, for players of Arena Of Valor. This gives them a better gaming experience, for which they pay a premium. The operator wants to expand service and ecosystem innovation, and part of that effort is the ongoing development of its next-generation operations support system (OSS). Unlike the other two large service providers in the country, China Unicom controls its network with a single, centralized OSS. Some of the elements of its evolving OSS are demonstrated in the Catalyst, such as the AI platforms and applications like intelligent self-healing capabilities. The ultimate aim is that the OSS will be AN-oriented, intelligent, centralized and cloud-based. Some of the control functions are already exposed to the upper layer by TM Forum’s Open APIs with more functions over time. The applications will be cloud-native or containerized, so that they can be easily and rapidly scaled, bearing in mind that China Unicom’s network is huge, involving more than two million base stations and 10 million network equipment nodes.

Lei Lei comments, “There is tons of work to make the network work, and in making it work like a charm. We need to employ AI and IT applications to manage the network more intelligently.”

AN initiative

In 2019, TM Forum launched the Network Automation (AN) initiative with a white paper, Autonomous networks: Empowering Digital Transformation for the telecoms industry and China Unicom adopted the AN best practices and architecture for preliminary trials. Early results show AI applications could reduce OpEx by more than 50 million RMB ($7.4 million) annually, so the CSP decided on a proof of concept to explore the potential further and promote AN’s advantages. Lei Lei notes, “The AN architecture proposed by TM Forum is great and very practical: Using three layers and a fully closed loop is efficient.” He adds that the biggest achievement of this inaugural phase of the Catalyst is showing that the AI technology is mature enough to be employed in the network OSS and says it could decrease network alarms by up to 60% in specific scenarios. Already China Unicom has been able to reduce workloads to recover base station malfunctions by 8% to 12% because the network can self-heal and be active again very quickly using AI to predict and solve problems. The use case in the Catalyst was catering to the scenario of many people now working from home and generating huge amounts of traffic in residential areas rather than city centers. Lei Lei says AI will help provide better control of scaling up the network where necessary as well as scaling it down to save power when appropriate.

The hardest part

He says, “The most difficult part is in service innovation – we still lack a deep and broad understanding of the service innovations proposed by 5G and IoT, and next dimension networks. We are evolving to 5G and even 6G, but are not completely clear about how we can make more money from the networks. We need business innovation – a new landscape in which our to-be customers or subscribers are willing to pay for our technology innovations.

“How can we affordably make our services faster, more reliable and more profitable at the same time, and not only for the telcos, but for our ecosystem partners? Making enough money to fund the entire ecosystem is the most difficult part. With autonomous networks, saving money is not the point; earning money is.”

The team is looking to the next phase of the Catalyst project, which will be demonstrated in summer 2021 if it goes ahead as planned.