Shulou(Shulou.com)11/24 Report--

Foreword:

With the vigorous development of 5G, high bandwidth, large connection, low delay, high reliable network capacity and big data, artificial intelligence continue to integrate, enabling the digital transformation of thousands of industries.

At the same time, the network itself, as the source of this change, is also carrying out its own digital transformation. This is the development goal put forward by the major operators recently: self-intelligent network.

In this issue, mayday will discuss those things about the self-intelligence network with you.

one。 Wireless communication, which relies on artificial complex systems, is a glorious network made up of thousands of base stations, but maintaining the link between these thousands of base stations is an inefficient manual configuration.

At the beginning of my career, I was puzzled by the dwindling 2G network:

Complex wireless communication systems have a large number of parameter settings, which need to be manually planned, configured and optimized, and the efficiency is really low!

Handoff as the core of mobility management, the network does not even know which base stations can be used as handover target sets, and neighbors must be manually configured.

If you want the base station to save power, you have to go to a lot of trouble to analyze half-day traffic and network indicators, and then try it on a small scale with the most conservative parameters. For fear of saving electricity but allowing users to access the Internet slowly!

At that time, the network, like an extremely complex machine, the operation of all functions were completely dependent on human instructions.

Manual work, in high-intensity, monotonous and repetitive work, is inefficient and error-prone. As a result, we can only find the best solution in the blind touch the elephant, and all operations are like treading on thin ice, as if in the abyss.

Network operation and maintenance needs to break the manual shackles and introduce automatic and intelligent operation in order to break the situation.

two。 The introduction of self-organizing network finally, in the 4G era, the concept of SON (Self Organized Network) is introduced. It is hoped that the operation of the network can reduce human intervention as much as possible and operate independently through self-organization.

This is a series of schemes through protocol standardization, which can realize the three major functions of site self-configuration, self-optimization and self-healing.

Self-configuration: the base station can be quickly opened, plug and play, including transmission self-establishment, parameter self-configuration, system self-test, asset self-reporting and a series of functions.

Self-optimization: the network can optimize itself according to user distribution, behavior and traffic, including automatic neighborhood relations, automatic load balancing, mobile robustness optimization, random access optimization and so on.

Self-healing: if a base station fails, you can automatically identify and adjust the parameters of other nearby base stations so that the coverage of these base stations can be extended to eliminate the coverage holes caused by the faulty site, and then restart the faulty base station to see if it can be repaired.

SON is like a ray of light, freeing valuable manpower from inefficient parameter configuration, optimization and performance improvement, and opens the prelude to network management automation.

Subsequently, 2G and 3G networks also introduced a series of similar SON functions, which takes the attempt of network management automation a step further.

However, human inertia is endless, with the SON function of wireless network, still feel almost meaningful.

Let's first briefly introduce what the network operation and maintenance is, and then talk about the shortcomings of SON.

The life cycle of wireless network is generally divided into planning, construction, maintenance, optimization and operation.

Planning: what is the goal of network construction, what kind of coverage and capacity requirements to achieve, where and how many sites need to be built?

Construction: how to make the site planned on paper become a signal-capable entity that can support users to make phone calls and surf the Internet? This requires a series of links such as site installation, debugging, parameter configuration, acceptance and so on.

Maintenance: after the network is built, there will inevitably be a variety of problems, such as internal hardware damage, system failures and so on, as well as external power outages, natural disasters and so on, which need to be repaired through network maintenance.

Moreover, the network is not immutable, and it may need to adjust configuration, expand capacity, introduce new features, upgrade versions and other operations from time to time, which also need network operation and maintenance.

Optimization: adjust the engineering parameters and network parameters of the site according to the traffic model of the specific network, so as to optimize the capacity, coverage and performance of the network.

Operation: operators spend a lot of money to build a network, how to evaluate user perception, how to improve ARPU, how to deal with user complaints, how to prevent users with poor experience from "voting with their feet"? These are all things that network operators need to do.

In general, what we often call "network operation and maintenance" can be understood in a narrow and broad sense. In the narrow sense, the network operation and maintenance refers to the network maintenance mentioned above, while in the broad sense, the network operation and maintenance refers to the whole process of "planning and maintenance".

Well, with the above groundwork, when we look at the whole operation and maintenance process of "planning, maintenance and maintenance camp", SON's set of "self-configuration, self-optimization, and self-healing" solves a few issues in "construction, maintenance, and optimization", and does not involve "regulation" and "camp" at all.

This is obviously not enough.

Today's network 2345G has four generations living together, high, middle and low frequency bands are varied, ToB / ToC business needs are different, the capabilities of various terminals are different, and the difficulty of operation and maintenance is increasing exponentially.

In the face of such a complex behemoth, only introducing "intelligence" in the whole life cycle and making the network "autonomous" is the fundamental solution.

three。 Self-intelligent network evolution 5G, since 2019 began to break the ground, ushered in the official commercial and rapid development.

In the same year, TM Forum (TMF) launched the initiative and established the "Autonomous Networks (literal translation: autonomous network)" collaboration project to explore and provide industry-leading end-to-end network automation methodology, and jointly with a number of enterprises in the industry released the "Autonomous Networks White Paper 1.0".

Note that I use "Autonomous Networks" directly here because there is no Chinese version of the white paper released at the beginning. Because the white paper establishes the six-level automation level of network operation and maintenance with reference to the classification of self-driving cars, it is translated as "self-driving network" in Chinese.

In addition, GSMA, 3GPP, ETSI, IETF, CCSA and other organizations have also expounded the goal and vision of network autonomy in the form of white papers or research reports, and all parties join hands to open the prelude of network automation, standardization and industrialization.

In 2020, TMF's "self-driving Network White Paper 2.0" was released. A number of equipment manufacturers, including Huawei, have also issued their own white papers on self-driving network solutions.

Personally, I think the name of self-driving network is very bad, and it is easy to make people think that giants such as China Mobile and Huawei have misled a lot of people when they want to enter the field of self-driving.

In 2021, TMF listened to the opinions of various experts and officially released the Chinese version of the white paper, and the official Chinese name of "Autonomous Networks" was also officially defined as "self-intelligent network", which is the "self-intelligent network white paper 3.0".

After several years of technical standardization and solution exploration, the industry has reached a consensus on the general direction of promoting network automation and intelligence, and has become the goal of the whole industry chain.

The vision of ICT is to provide a good customer experience for vertical and consumer users through a fully automated network and ICT's intelligent infrastructure, agile operations and panoramic services, using cutting-edge technology to achieve "leaving complexity to suppliers and minimalism to customers".

The network mentioned here is an end-to-end concept, including access, transmission, core network, data center and other fields, using the same set of methodology to make multiple domains work together. In a word, it is to make the whole network automated and intelligent, so as to provide users with a better service experience.

To sum up, the self-intelligent network is to realize "303 self".

The "Zero-X" in this means: zero wait, zero contact, zero failure.

Zero waiting: the release, delivery and maintenance of various services on the network are very fast, and there is almost no need to wait.

Zero contact: all kinds of processes can be automated with minimal or no human intervention.

Zero failure: the network has a strong self-healing ability, the outside world can not feel the existence of the fault.

To achieve the above "three zeros", the network needs to have "three self (Self-X)" capabilities: self-service, self-distribution, self-protection. The specific description is shown in the following figure.

Different equipment vendors and operators can also give different explanations of "303 self" to different network domains according to their own plans.

The framework of self-intelligent network consists of "three levels" and "four closed loops".

The "three levels" include resource operation layer, service operation layer and business operation layer, which can support all scenarios and business requirements as a general operational capability.

Resource operation layer: mainly provides network resources and capability automation for a single autonomous domain. For wireless communication, wireless access network (RAN) can be regarded as an autonomous domain, which together with other autonomous domains such as transmission and core network constitute the resource layer.

Service operation layer: mainly provides IT services, network planning, design, launch, distribution, guarantee and optimization of operation capabilities for multiple autonomous domains.

Business operation layer: mainly for self-intelligent network business, providing the enabling and operational capabilities of customers, ecology and partners.

These three levels cooperate with each other to form the "four closed loops", including:

Resource closed loop: the interaction between the network and ICT resource operation with the granularity of autonomous domain. To put it simply, it is: single-domain autonomy and cross-domain cooperation. Taking the wireless access network as an example, the closed loop of resources requires the automation and intelligence of the whole process of "planning, construction and maintenance".

Service closed loop: the closed loop between the service, network, and IT resource operation layer. The service closed loop may trigger the associated resource closed loop in its implementation.

Business closed loop: the interaction between the business and service operation layers. The business closed loop may invoke the relevant service closed loop and resource closed loop in its implementation.

User closed loop: the interaction between the above three levels and the other three closed loops to support the implementation of user services. The three levels interact with each other through an intention-driven minimalist API interface.

Intention-driven interaction in the autonomous network, the information interaction between different autonomous domains and different levels is based on "intention-driven".

So, what is the intention? To put it simply, it is goal-oriented.

For example, when leaders assign tasks to their subordinates, they usually only say what work they want to do and what results they hope to achieve, but they will not list detailed and specific steps one by one. In other words, as long as the goal can be achieved, the process can be played freely.

In the self-intelligent network, the most basic unit is the autonomous domain. Now that people are autonomous, they must not be able to teach people how to do things by hand, as long as they can convey their intentions. Exactly how to do it, the autonomous domain will take care of it on its own.

In this way, we cooperate with each other, not only reduce the technical complexity, but also shield the differences between different vendors, so as to support the end-to-end business requirements of the self-intelligent network.

Take the wireless access network as an example, if we want to plan indoor coverage in a building, we don't have to do any specific work, we just need to say to the system (literally, machines can recognize speech and understand natural language):

"how should users in this building build a network if they want to achieve a good user experience?"

The machine will comprehend your intention by itself, and then, after analysis, calculation, and decision-making, it will give the best plan: whether to build a traditional room or a digital room, how many rooms are needed, where they should be placed, and what network indicators can be achieved after they are built.

Isn't it convenient? People interact with machines in this way, and machines and machines interact in the same way. This is the goal of the intelligent network.

The classification of self-intelligent network refers to the classification of autopilot, and the level of self-intelligent network is also divided into six levels from L0 to L5, and the degree of automation of the network is enhanced in turn.

L0: manual operation and maintenance-the system provides auxiliary monitoring capabilities, and all dynamic tasks need to be performed manually.

English: auxiliary operation and maintenance-the system can perform specific repetitive subtasks according to pre-configuration to improve execution efficiency.

L2: partial self-intelligent network-in a specific external environment, the system can automate closed-loop operation and maintenance for specific units according to predefined rules / policies.

L3: conditional self-intelligent network-on the basis of L2, the system can perceive environmental changes in real time and make self-optimization and self-adjustment in specific network majors to adapt to the external environment.

L4: highly self-intelligent network-based on L3, the system can analyze and make decisions by implementing predictive or proactive closed-loop management of business and customer experience-driven networks in a more complex multi-network environment.

L5: fully self-intelligent network-this level is the ultimate goal of telecom network evolution, and the system has the ability of full-scene closed-loop autonomy for multi-service, multi-domain and full life cycle.

At present, the vast majority of operator networks have a self-intelligence level of L1 or L2. Many operators have set the goal of achieving high-level self-intelligent networks, and L5 is still out of reach, so L4 has become a common goal.

four。 At the end of the network automation will greatly improve the efficiency of operators and enable new digital services, while enabling new services, change society and thousands of industries. This has become the consensus of the industry.

At present, the self-intelligent network is in the critical stage from concept to implementation, and many applications are still in the initial stage, showing point-like development. With the joint efforts of the whole industry, the waves of these self-intelligent networks are bound to set off huge waves of network intelligence.

Because, the self-intelligent network is no longer an option, but the only way.

This article comes from the official account of Wechat: wireless Deep Sea (ID:wuxian_shenhai), author: mayfly acquisition

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.