Joint innovation drives China’s ‘big three’ toward autonomous networking

• Who: China Mobile, China Unicom, and China Telecom with Huawei Technologies Co., Ltd
• What: Driving the long-term shift to autonomous networking through joint innovation
• How: Huawei has created a lab for AI innovations for the telco industry, in which AI-based KPIs are developed using TM Forum’s AI & Digital Maturity Model, and applied in its AI-driven solutions to help China’s top three operators automating certain manual processes in various domains and realize benefits on their path to the autonomous network. The ongoing work led to winning Catalyst projects for AIOps two years running and led to the publication of the AIOps Service Management deliverable
• Results: Huawei showed that operators can realize benefits in multiple domains and at each stage on the long road to the autonomous network. China Mobile, China Unicom and China Telecom all benefited from improvements in the areas to which they applied automation and AI solutions. For example, China Mobile realized an alarm compression rate of 99%, a 6% improvement in RF resource utilization for 5G. China Unicom decreased packet loss in its VoLTE service by 5% and China Telecom reduced energy consumption by 10%.

Upending the balance of escalating cost and diminishing revenue

CSPs hardly continue to operate in traditional ways and remain profitable or competitive. Manual processes and labor-intensive methods for managing and optimizing the network are not sustainable. CSPs need to move away from the passive management of network device-centric networks, which gets more cost-prohibitive every day. Revenue is not increasing commensurate with the traffic volume CSPs are required to support. 5G will only increase the disparity between growing cost and shrinking revenue. Also, cost and revenue aside, it is becoming physically impossible to manage increasingly complex and high-volume networks manually. Structural innovation for network operations is now required. Huawei is working with CSP partners, to move towards autonomous networks that are proactive, intelligent, accurate and automated in a way that solves these challenges and does so in a way that incrementally improves both efficiency and profit margins. It also allows CSPs to free their operations staff from repetitive functions that machines can do more efficiently and accurately to engage in high-value activities such as service innovation and customer care.

A roadmap to automation

Because the road to autonomous networking is long and arduous, the telco industry needs to get started, especially if they can demonstrate the benefits at each stage, rather than hoping to see benefits at the end of the project. With a long-range goal of achieving a fully autonomous network end-to-end by 2030, Huawei worked jointly with China Mobile, China Telecom, and China Unicom, to develop and trial in live networks the Autonomous Driving Network Solution Working with multiple operators allowed the teams to work on diverse network domains, including wireless access, core, IP, optical transmission, and fixed access networks. Huawei followed the approach outlined in TM Forum’s Autonomous Network Whitepaper to start the process from a single technology domain within each CSP that pragmatically targets their CSPs’ key pain points. Within each domain Huawei enabled the CSPs to offer zero-wait, zero-touch, zero-trouble services, which means they were near real-time, automated and worked as configured. Huawei also helped them maximize network asset utilization using full lifecycle automation. Some operators worked on the same domains within their business, while others had specific needs.

A brief description of each case study is below:

China Mobile – Huawei and China Mobile found five key pain points that artificial intelligence (AI) and automation could relieve now on the CSPs path to an autonomous network:

China Unicom found three key pain points to which Huawei’s AI and automation solutions could apply:

China Telecom focused on energy efficiency and anomaly detection:

• Alarm aggregation – Overwhelmed by the 600,000 packet transport alarms it received daily, China Mobile needed a new way to address troubleshooting the network. It had to transform from an alarm-centric operation to an incident-centric operation. Using AI, Huawei helped the operator compress alarms by up to 99% to approximately 600 incidents.
• Resource utilization in 5G RF propagation – China Mobile also addressed resource utilization in its 5G network by automating massive multiple input, multiple output (MIMO) optimization. RF utilization improved by 6% in the 5G network while the aggregate traffic for the MIMO cell increased by 14.2%
• Energy efficiency – Huawei introduced its AI-based PowerStar smart energy solution in 2018, China Mobile used it to reduce its energy consumption by 10%. At this rate of reduction, a typical wireless network of any generation can prevent two million kilograms of carbon dioxide emissions from entering the atmosphere annually. This solution allows networks to fit custom power-saving strategies with configuration and traffic needs on different bands and modes at the base station level. With these strategies, mobile users can be switched to lower bands when total traffic remains low so that high bands can be switched off to realize deep power saving.
• Optical network troubleshooting – Huawei applied its premium home broadband solution to topology visualization and remote fault analysis to reduce China Mobile’s home visit rate by 20% and improves troubleshooting efficiency by 25%.
• Throughput optimization -- Through iterative optimization that continually improved AI algorithms for better multi-carrier performance optimization, China Mobile showed with road tests that throughput improved by 14.5% by automatically selecting faster Internet speeds for consumers with the same hardware.
• Throughput optimization – Like China Mobile, China Unicom leverage iterative AI optimization to improve its cell throughput. The operator increased throughput by 15%.
• VoLTE packet loss – Packet loss, according to a 2019 study by Empirix and the University of Modena and Reggio Emilia in Modena, Italy, is the most significant influence the end-to-end quality of VoLTE calls, affecting the intelligibility and naturalness of the reconstructed speech. Huawei and China Unicom applied AI-based parameter optimization and decreased the VoLTE packet loss rate by 5%. They also reduced fault isolation times to 1 minute.
• 5G RAN self-healing – Using AI-empowered intelligent operation applications for 5G, including RAN self-healing and autonomous optimization, helped decrease by 30% the workload on network maintenance.
• Energy efficiency – China Telecom joined China Mobile with a focus on energy savings and also reduced its consumption by 10% by intelligently shutting down unnecessary cells, frequency bands, or chipsets.
• Anomaly detection – In 5G when many devices can suddenly attempt to connect at the same time, or when the occurrence has more malicious origins, the ability to detect signaling storms is critical for efficient operations and quality experience. The operator used AI-based KPI Anomaly Detection algorithms to get an early warning for core network signaling storms, achieving 71% accuracy. In a Heavy Reading study in 2019, 68% of respondents believe there is a greater need to implement signaling protection against protocol attacks.

Where to begin the journey

The most immediate challenge for CSPs transforming into an autonomous network is picking a place to start. Since that starting point will likely be different for each operator, Huawei created its Automatic Driving Network (ADN) solution to work in any operating environment with varying levels of automation, different organizational structure, processes and operational and business support systems (OSS/BSS). Thus, ADN has two layers. One for simplified networks and one for intelligent operations. Both focus on easily duplicating deployments across environments. Another key challenge across the industry is the availability of data with which to leverage AI and other analytics tools.

CSPs are reluctant to share customers and some performance data which could be used to continuously improve AI systems. In response, Huawei has built a large-scale lab for simulating and generating data for initial AI model training. These models are verified in live networks, and the experience and knowledge are cycled back into the lab to generate new, improved models. Huawei also applied its iMaster network automation and intelligence platform to integrate the management, control, analysis, and AI functions that provide centralized management, control, and analysis of CSP networks. The products within this platform include that Huawei applied to these transformations include: Huawei Network AI Engine (iMaster NAIE) and the Autonomous Network Management and Control System (iMaster NCE). Huawei also applied its Microwave Wireless Network and Digitized Operations Services (AUTIN – short for automation and intelligence).

IMaster NAIE for high-precision model training

Huawei innovations and technology also worked in conjunction with TM Forum assets to help CSPs get started on the path toward automation. As mentioned above, Huawei used the AI Maturity Model as a reference framework to develop its AI-based KPIs, an essential starting point. Huawei demonstrated many of these capabilities in the award-winning AIOps Catalyst project. The AIOps Catalyst project has completed its third phase with participation from 12 companies. The knowledge, insights and experience drawn from these Catalysts have been invaluable in contributing to the advancement of AI-based solutions. The Catalyst team has now developed eight use cases addressing the various business needs presented by the CSP champions. These cut across customer experience, quality of service, business performance and efficiency.

The work has been contributed to the publication of TM Forum’s AIOps Service Management document. China Unicom’s AI PaaS is also fully compliant with TM Forum Open APIs. Huawei and its CSP partners derived their methods in part from the Autonomous Networks whitepaper, to which they also contributed and which became the impetus for the Autonomous Networks Collaboration Initiative China Mobile, China Unicom, China Telecom and Huawei will continue to make business and technical contributions to develop an industry-wide common understanding and consensus on the autonomous network concept and automation classification for the simplification of telecom network infrastructure, automated & intelligent operations and innovative services.