During the fifth Global Ultra Broadband Summit Forum, Huawei officially put forward the “self-driving network” for the first time. Then, when many people first encounter the concept of autonomous driving networks, they understand it as — how did Huawei come up with cars?
A self-driving Network is a kind of Network that is predictable and capable of running autonomously. Simply understand Huawei or do IT infrastructure network products, but is “network” autopilot. What is the autopilot of the network? This question, everyone has their own understanding, next I will answer this question according to their own understanding.
1. Understand the autonomous driving network from a macro perspective
The self-driving network name is borrowed from the concept of self-driving cars, as it is commonly known in the industry. Although it is called Autonomous Driving Network in Chinese, the English equivalent is Autonomous Driving Network. Some people also use self-driving Network, while TMF uses the name of Autonomous Network. The word “Autonomous” means Autonomous, emphasizing the network’s ability to make Autonomous decisions. However, the word autonomy is actually a little inappropriate after careful consideration, because the network serves people, it must be carried out according to people’s ideas, how can autonomy come? There’s a similar debate about what to call Autonomous vehicles, and it’s interesting that at the end of the day Autonomous vehicles in English are still referred to as Autonomous. I don’t think it’s a big problem, because it’s more like a brand. What is a brand? Brand is a symbol, which is endowed with a value proposition. For example, the story of Nike shoes’ trademark tells that the curved hook is like the wings of Greek goddess of victory, representing speed, movement and gentleness. This proposition is very consistent with Nike’s characteristics, whether it is true or not is not important. General brand easy to remember, positive association is good.
From an implementation perspective, what is the value proposition of autonomous driving? What is the specific connotation? This is more instructive for the rest of the work. Regardless of the path, let’s imagine what a self-driving network would look like.
I believe Ren Zhengfei’s judgment that “human society will inevitably enter an intelligent society in the next 20 or 30 years”. Artificial intelligence is now in a period of rapid development, various algorithms and functions are dazzling, but the intelligent society is not just a change brought by a few intelligent algorithms, but a comprehensive change of social infrastructure. The network is the enabler of the change of social infrastructure, and to realize this enabling function, the supporting form of the network must be different from the existing model. “When the whole society becomes more complex, it is advantageous to provide simpler and simpler networks,” Ren said.
The vision of the autonomous driving network is a self-subversion of the current network. According to Ren Zhengfei’s understanding of the network supporting an intelligent society, the autonomous driving network should be a new attempt to drive the progress of the industry and promote the progress of human society. So the self-driving network is not simply from the network itself to understand.
In modern society, there is a problem called baumol’s cost disease. It refers to the problem that labor productivity has increased greatly in most manufacturing sectors since the Industrial Revolution, but it is difficult to improve labor productivity in the service sector. Efficiency gains are most evident in electronics, where many products are now better and cheaper than ever before and manufacturers can still make money. It is much harder to improve the efficiency of services: finding private tutors for children, for example, is now more expensive than it used to be. However, this situation will be greatly changed with the emergence of intelligent platform. In the past, the efficiency of taxi is very low. Many taxis have to sweep the street in order to find customers, and run around empty to try their luck. But with the digital intelligent platform for taxi-hailing, intelligent scheduling of vehicles can be carried out, and the efficiency of taxis has been greatly improved. The telecom industry also has a similar problem, network products are more and more advanced, but the way of network operation and the ability to provide services to the outside is still very traditional, the efficiency is very low, how to change this situation is the autonomous driving network needs to answer.
We are now more familiar with the network automatic driving function is self-configuration, self-optimization, self-healing several characteristics. However, these features were introduced in the development of 4G SON AD hoc networks. While the SON concept is clear, there seems to be nothing special about autonomous driving networks, if only based on the above features. However, if we look at the future from the day after tomorrow, the biggest difference between SON and AUTONOMOUS driving network should be that SON defines characteristics from the perspective of product, while autonomous driving network should define characteristics from the perspective of network as an enabler of infrastructure. Although the functions realized by the network may still be self-configuration, self-optimization and self-healing, the system design will be completely different because of the different perspectives.
The most traditional network is to provide a telephone function, the user is insensitive, there is no need for customization, marketing is simple, system support is optional. But for the future network industry applications, demand is multifarious, alone a 5 g slice function is likely to involve a lot of network scheduling, demand is likely to change at any time, and control network users from simple operations staff extend to all kinds of users, including users and operations staff, developers and the relevant ecological personnel, marketing personnel. At this time, the original operation mode will certainly be difficult to use. For example, many functions are set by operation and maintenance personnel, so why can’t users directly control network functions? All of these new operational requirements should be met by autonomous networks.
From the user’s point of view, it may drive changes to things that are routine. Such as I speak to innovation course, specifically, to experience the tesla motors, a detail is the car is got on the car to press start switch before starting the car, I found on the tesla along while also didn’t find the start switch, originally the tesla didn’t this switch, go get on the car into gear, tesla still has a lot of change is refreshing, An unexpected but logical change, this is the result of defining features from a user perspective. Behind in terms of how network automated driving, I will have a story about how to change the method to realize the fault handling, fault handling are now warning as the center, everybody is thinking about how to reduce the alarm, but we deal with failure, why must deal with so many alarm and not deal directly with the fault information? This approach can be fully realized in the age of intelligence, and there will be a lot of changes in the future network, so there are still a lot of things to be done in the autonomous driving network.
Although e-commerce platform, online car booking platform, food ordering platform and other intelligent digital platform operation is very successful, but once entered into the specific manufacturing, electricity, banking, telecommunications and other industries, the platform is not easy to succeed, there are a lot of industry know-how, there are many professional operation technology OT. The platform must get through the whole link of technology, business, data and process, which becomes very difficult. Telecom industry has accumulated a lot of practical experience, the history has a lot of practical system, but is the chimney system, change is difficult, so automated driving small refinements of the network is certainly not the current network, more should from the frame has a new vision, to integrate the network itself and management ability, integrated together to build a new overall solution, so just have the largest collaborative.
2. Future autonomous driving network core capabilities
Although the implementation of specific functions of autonomous driving networks can take many specific forms, the following functions are essential from the user’s point of view. When these functions are implemented, the network can maximize the convenience of users and do the least to maintain itself.
For example, in the future, the network is required to be uninterruptible, but the hardware will always fail. At this time, the network is required to have several capabilities:
- The first is perceptibility, the ability to know that something is going wrong.
- The second is that resources must be pooled and uniformly scheduled, so that there may be redundant resources to avoid the point of failure.
- The third is intelligent rerouting in network scheduling.
These three functions are essential for automatic fault recovery. These three functions also require close cooperation between the network and network management. Furthermore, if it is connected with the availability management of the network, the availability management process should closely cooperate with the above three functions, and finally achieve the optimal state of the network and the minimum involvement of personnel. Other functions are also necessary. For example, if the network needs to adapt to a variety of application environments, it is not enough to simply rely on the customized functions of operators. Therefore, it is necessary to provide open interface functions to meet the needs of various industries. In order to meet the needs of network analysis, the intelligent analysis function of network management is essential.
3 Phased path of realization
At present, the autonomous driving network imitates autonomous driving vehicles and is divided into 5 levels. However, due to the characteristics of the network, the definition of network level directly copied is difficult to understand and difficult to operate in practice. According to the actual development path of the possible industry, I set phased goals for network management and network itself, and also divided into five levels. At present, because the ability of automatic recovery of network products is still relatively weak, and the network is still at the recommendation level, most of the functions implemented should focus on the improvement of network awareness, such as automatic fault discovery, fault cause recommendation and other functions. When the future network product capability is enhanced and the network has the function of quality perception and rerouting, the capability of network implementation becomes the control level. Finally, the network can access a variety of environments, complete the support of a variety of services, the network will achieve a perfect automatic driving network function.
4 SMART² Architecture SMART 6S Full network view
In order to realize the core network functions above, I made a schematic diagram of an architecture that conforms to the principles of SMART2 (Service, multi-access, AI, Robust, highly available, Trust and Tailored). SMART² is a pun on Tailored networks, which are Tailored for SAMRT and SMART network management. SMART² means extremely SMART. In terms of usage requirements, autonomy is not the main feature of the future network, but the ability to execute intelligently according to people’s intention is the most important feature.
If I had the freedom to choose the name of the future network, I would call this kind of network in Chinese intelligent simple network. The network must be Smart, and the result of Smart is to bring minimalist network transmission and minimalist operation and maintenance management.
We used to be a minimalist network, the simplicity of network switching and transmission, and now we’re going from the simplicity of product networks to the intelligence of the whole network, where you can simplify your work. Intelligence is not meaningful in itself, only if it is ultimately simple to use. Wise Jane, to Jane, with wisdom to Jane.
If the name is English, I will use 6S network, in addition to product network and network management SMART and SIMPLIFIED, network also needs SAFETY and SECURITY, a total of six words starting with S. Just cover the key features of the future web.
The above is my analysis of the ins and out of the ins and out of the ins and out of the ins and out of the ins and out of the ins and out of the ins and out of the ins and out of the ins and out of the ins and out of the ins and out of the ins and out.
Click to follow, the first time to learn about Huawei cloud fresh technology ~
