Huawei releases next-generation O&M architecture for autonomous network operation, expected to save operators 25% of O&M costs

On April 17, 2019, at HAS2019, Huawei released the next-generation network architecture aimed at autonomous driving. It is expected that within 10 years, telecom operator networks will enter the L5 autonomous driving stage, saving operators 25% of operation and maintenance costs, said Mr. Han Yufa, President of Huawei SoftCOM AI Solutions.

[[262823]]

In recent years, the trend of network autonomous driving has gradually become a consensus of the entire telecommunications industry, said Mr. Han Yufa. Two major driving forces have driven the transformation of the network. The first is business demands. By 2025, the number of connections in the entire network will reach 75 billion. The network will have multiple access technologies coexisting to meet the connection demands. Fixed network access and 2G3G4G5G multi-domains will coexist, and the network complexity is more than 10 times that of the current level. The second is that the degree of automation of operator network operation and maintenance is not high. The current operator network still relies heavily on human experience and skills. Compared with OTT, operators need 384 people to operate and maintain a network with 10,000 devices, while OTT only needs 3 people. The result is that the operator's OPEX is increasing year by year (Growing OPEX): the proportion of OPEX is also increasing year by year. According to statistics, it is 3 to 4 times the CAPEX. On the one hand, the complexity of the network is growing exponentially, and on the other hand, the cost of network operation and maintenance remains high. Operators are already overwhelmed and the network calls for change. Currently, 63% of telecommunications organizations have begun to invest in AI autonomous driving, and all TOP big Ts have clarified the development strategy of autonomous driving. Telecom autonomous driving networks have become an industry consensus.

For network autonomous driving, Huawei has released the next-generation network operation and maintenance architecture. Based on a unified cloud platform, it is oriented to 5G network planning, agile private lines, predictive maintenance, intelligent tuning and other scenarios. It uses AI, Cloud, DigitalTwin and other technologies to achieve layered (cloud intelligence, network intelligence, edge intelligence) and domain (wireless, fixed, core network, data center) autonomy. The core of the autonomous driving network is business-oriented network agility, automation and intelligence. Compared with traditional networks, the operation and maintenance for network elements is transformed into scenario-oriented (planning, construction, maintenance optimization, business deployment) operation and maintenance, network services are deployed in minutes, network operation and maintenance realizes policy-based automated closed loop and fault self-healing, dynamic optimization of energy consumption, and on-demand use of resources. The ultimate goal is to double resource utilization, double energy utilization, double maintenance efficiency and improve end users. Mr. Han Yufa said that we are currently in the L2 network autonomous driving stage, which is human-centered and machine-assisted. In the next 10 years, Huawei's network autonomous driving architecture will help operators move from partial autonomous driving (L3) and highly autonomous driving (L4) to L5 comprehensive autonomous driving, achieving a 70% efficiency improvement and a 25% reduction in operation and maintenance costs.

As an enabling technology for network autonomous driving, Huawei's SoftCOM AI solution has gone from concept to reality. The core of the SoftCOM AI solution consists of four components: a cloud training platform, which is responsible for model training and converts expert experience into models; an inference framework, which translates real-time data into network control instructions based on models and sends them to the network management system for execution; MarketPlace, which is responsible for managing trained models and pushing them to the network; and a data lake, which helps operators build data warehouses. This solution is designed specifically for telecommunications systems, and Huawei's 30 years of network experience is embedded in it; the model is provided as a cloud service and is available everywhere in the operator's network; a one-stop development and deployment environment, where operators can get guidance on the entire process of developing and deploying models, and 90% of the work has been pre-set.

This solution has been working with multiple operators in the past year to solve their practical problems, said Mr. Han Yufa. Case 1: Operator base stations consume a lot of energy, and the ratio of traffic volume during busy and idle hours is dozens of times, while the ratio of base station energy consumption during busy and idle hours is only a few times. When the traffic volume is low at night, the base station radio frequency unit is still fully powered, which wastes energy seriously. A mobile operator uses Huawei's SoftCOM AI solution to accurately predict traffic flow and evaluate network KPI indicators at the same time. It puts idle carriers into a sleep state while ensuring that KPI indicators do not decrease. After deployment throughout the province, it saves 20 million kWh of electricity a year, saving 10% energy. Case 2: In the operator's network, a single fault will trigger fault alarms of multiple connected components. A provincial company generates 20 million alarms a day, and there are still 2 million after initial compression. Alarms will trigger work orders, and operators need to pay for massive work orders. Using Huawei's SoftCOM AI solution, machines can learn the correlation between fault alarms and find the root cause of the fault, thereby compressing work orders. After three months of operation, a city mobile company further reduced its work orders by 21.5%, and is expected to save 2 million yuan per year.

Mr. Han Yufa said that the next decade will be an era of network intelligence, and Huawei is willing to work with all partners who are committed to this goal to jointly promote network autonomous driving.