This World Cup, the “smart referee” is one of the biggest highlights. SAOT integrates stadium data, game rules and AI to automatically make quick and accurate judgments on offside situations
While thousands of fans cheered or lamented the 3-D animation replays, my thoughts followed the network cables and optical fibers behind the TV to the communications network.
In order to ensure a smoother, clearer viewing experience for fans, an intelligent revolution similar to SAOT is also under way in the communication network.
In 2025, L4 will Be Realized
The offside rule is complicated, and it is very difficult for the referee to make an accurate decision in a moment considering the complex and changeable conditions of the field. Therefore, controversial offside decisions frequently appear in football matches.
Similarly, communications networks are extremely complex systems, and relying on human methods to analyze, judge, repair, and optimize networks over the past few decades is both resource-intensive and prone to human error.
What is more difficult is that in the era of digital economy, as the communication network has become the base for the digital transformation of thousands of lines and businesses, the business needs have become more diversified and dynamic, and the stability, reliability and agility of the network are required to be higher, and the traditional operation mode of human labor and maintenance is more difficult to sustain.
An offside misjudgment may affect the result of the whole game, but for the communication network, a “misjudgment” may make the operator lose the rapidly changing market opportunity, force the production of enterprises to be interrupted, and even affect the whole process of social and economic development.
There is no choice. The network must be automated and intelligent. In this context, the world’s leading operators have sounded the horn of self-intelligent network. According to the tripartite report, 91% of global operators have included autointelligent networks in their strategic planning, and more than 10 head operators have announced their goal of achieving L4 by 2025.
Among them, China Mobile is in the vanguard of this change. In 2021, China Mobile released a white paper on self-intelligent network, proposing for the first time in the industry the quantitative goal of reaching level L4 self-intelligent network in 2025, proposing to build network operation and maintenance capability of “self-configuration, self-repair and self-optimization” inward, and create customer experience of “zero waiting, zero failure and zero contact” externally.
Internet self-intelligence similar to “Smart Referee”
SAOT is made up of cameras, in-ball sensors and AI systems. The cameras and sensors inside the ball collect the data in full, real time, while the AI system analyzes the data in real time and accurately calculates the position. The AI system also injects the rules of the game to automatically make offside calls according to the rules.
There are some similarities between network autointellectualization and SAOT implementation:
Firstly, network and perception should be deeply integrated to comprehensively and real-time collect network resources, configuration, service status, faults, logs and other information to provide rich data for AI training and reasoning. This is consistent with SAOT collecting data from cameras and sensors inside the ball.
Secondly, it is necessary to input a large amount of manual experience in obstacle removal and optimization, operation and maintenance manuals, specifications and other information into the AI system in a unified manner to complete automatic analysis, decision making and execution. It’s like SAOT feeding the offside rule into the AI system.
Moreover, since the communication network is composed of multiple domains, for example, the opening, blocking and optimization of any mobile service can only be completed through the end-to-end collaboration of multiple subdomains such as wireless access network, transmission network and core network, and network self-intelligence also needs “multi-domain collaboration”. This is similar to the fact that SAOT needs to collect video and sensor data from multiple dimensions to make more accurate decisions.
However, the communication network is much more complex than the football field environment, and the business scenario is not a single “offside penalty”, but extremely diversified and dynamic. In addition to the above three similarities, the following factors should be taken into consideration when the network moves towards higher-order autointelligence:
First, the cloud, network and NE devices need to be integrated with AI. The cloud collects massive data across the whole domain, continuously conducts AI training and model generation, and delivers AI models to the network layer and NE devices; The network layer has medium training and reasoning ability, which can realize closed-loop automation in a single domain. Nes can analyze and make decisions close to data sources, ensuring real-time troubleshooting and service optimization.
Second, unified standards and industrial coordination. Self-intelligent network is a complex system engineering, involving many equipment, network management and software, and many suppliers, and it is difficult to interface docking, cross-domain communication and other problems. Meanwhile, many organizations, such as TM Forum, 3GPP, ITU and CCSA, are promoting self-intelligent network standards, and there is a certain fragmentation problem in the formulation of standards. It is also important for industries to work together to establish unified and open standards such as architecture, interface and evaluation system.
Third, talent transformation. Self-intelligent network is not only a technological change, but also a change of talent, culture and organizational structure, which requires the operation and maintenance work to be transformed from “network centered” to “business centered”, operation and maintenance personnel to transform from hardware culture to software culture, and from repetitive labor to creative labor.
L3 is on its way
Where is the Autointelligence network today? How close are we to L4? The answer may be found in three landing cases introduced by Lu Hongju, president of Huawei Public Development, in his speech at the China Mobile Global Partner Conference 2022.
Network maintenance engineers all know that the home wide network is the biggest pain point of the operator’s operation and maintenance operation work, perhaps no one. It is composed of home network, ODN network, bearer network and other domains. The network is complex, and there are many passive dumb devices. There are always problems such as insensitive service perception, slow response, and difficult troubleshooting.
In view of these pain points, China Mobile has cooperated with Huawei in Henan, Guangdong, Zhejiang and other provinces. In terms of improving broadband services, based on the collaboration of intelligent hardware and quality center, it has realized accurate perception of user experience and accurate positioning of poor quality problems. The improvement rate of poor quality users has been increased to 83%, and the marketing success rate of FTTR, Gigabit and other businesses has been increased from 3% to 10%. In terms of optical network obstacle removal, the intelligent identification of hidden dangers along the same route is realized by extracting the optical fiber scattering characteristic information and AI model, with an accuracy of 97%.
In the context of green and efficient development, network energy saving is the main direction of the current operators. However, due to the complex wireless network structure, overlapping and cross-covering of multi-frequency band and multi-standard, the cell business in different scenarios fluctuates greatly with time. Therefore, it is impossible to rely on artificial method for accurate energy-saving shutdown.
In the face of challenges, the two sides worked together in Anhui, Yunnan, Henan and other provinces at the network management layer and the network element layer to reduce the average energy consumption of a single station by 10% without affecting the network performance and user experience. The network management layer formulates and delivers energy saving strategies based on the multi-dimensional data of the whole network. The NE layer senses and predicts the business changes in the cell in real time, and accurately implements energy saving strategies such as carrier and symbol shutdown.
It is not difficult to see from the above cases that, just like the “intelligent referee” in the football match, the communication network is gradually realizing self-intelligentification from specific scenes and single autonomous region through “perception fusion”, “AI brain” and “multi-dimensional collaboration”, so that the road to advanced self-intelligentification of the network becomes increasingly clear.
According to TM Forum, L3 self-intelligent networks “can sense changes in the environment in real time and self-optimize and self-adjust within specific network specialties,” while L4 “enables predictive or active closed-loop management of business and customer experience-driven networks in more complex environments across multiple network domains.” Obviously, the autointelligent network is approaching or achieving level L3 at present.
All three wheels headed for L4
So how do we accelerate the autointellectual network to L4? Lu Hongjiu said Huawei is helping China Mobile reach its goal of L4 by 2025 through a three-way approach of single-domain autonomy, cross-domain collaboration and industrial cooperation.
In the aspect of single-domain autonomy, firstly, NE devices are integrated with perception and computing. On the one hand, innovative technologies such as optical iris and real-time sensing devices are introduced to realize passive and millisecond level perception. On the other hand, low-power computing and stream computing technologies are integrated to realize intelligent NE devices.
Secondly, the network control layer with AI brain can combine with intelligent network element devices to realize the closed-loop of perception, analysis, decision making and execution, so as to realize the autonomous closed-loop of self-configuration, self-repair and self-optimization oriented to network operation, fault handling and network optimization in a single domain.
In addition, the network management layer provides an open northbound interface to the upper-layer service management layer to facilitate cross-domain collaboration and service security.
In terms of cross-domain collaboration, Huawei emphasizes the comprehensive realization of platform evolution, business process optimization and personnel transformation.
The platform has evolved from a smokestack support system to a self-intelligent platform integrating global data and expert experience. Business process from the past oriented to network, work order driven process, to experience oriented, zero contact process transformation; In terms of personnel transformation, by building a low-code development system and atomic encapsulation of operation and maintenance capabilities and network capabilities, the threshold of CT personnel’s transformation to digital intelligence was lowered, and the operation and maintenance team was helped to transform to DICT compound talents.
In addition, Huawei is promoting the collaboration of multiple standard organizations to achieve unified standards for self-intelligent network architecture, interface, classification, evaluation and other aspects. Promote the prosperity of industrial ecology by sharing practical experience, promoting tripartite evaluation and certification, and building industrial platforms; And cooperate with China Mobile smart operation and maintenance sub-chain to sort out and tackle root technology together to ensure the root technology is independent and controllable.
According to the key elements of the self-intelligent network mentioned above, in the author’s opinion, the “troika” of Huawei has the structure, technology, cooperation, standards, talents, comprehensive coverage and precise force, which is worth looking forward to.
Self-intelligent network is the telecommunications industry’s best wish, known as “telecommunications industry poetry and distance”. It has also been labeled as “long road” and “full of challenges” due to the huge and complex communications network and business. But judging from these landing cases and the troika’s ability to sustain it, we can see that poetry is no longer proud, and not too far away. With the concerted efforts of the telecommunications industry, it is increasingly full of fireworks.
Post time: Dec-19-2022