Author: Li Zhengmao
From the primitive society of drinking blood and eating meat to the intelligent society of intelligent association of all things, with the development of technology and human demand for a better life, productivity has developed rapidly. As we all know, productivity is composed of labor materials, labor objects and workers, in which people are the most active factor of productivity, and technology is the most key factor to promote the reform of productivity. Technology can enrich the sources and utilization methods of means of production, create and improve production tools, make them process and utilize the means of production effectively, and improve the skills of workers to find the value of means of production and use production tools. The development process of human social productive forces is actually an evolution process from simple human centered to technology centered. Specifically, the development of human productivity has gone through the evolution of human, animal and power. At present, it has entered the era of computing power, and computing power has become the core productivity.
Every major change in human history is accompanied by quadrant level technological breakthroughs. The invention of batch planting technology has brought mankind into the agricultural era; The invention of steam engine and thermal weapon has brought mankind into the era of industry and navigation; The invention of computer and Internet has brought mankind into the information age. The level of information computing and storage facilities can not be divided into three levels, and the power of information computing and storage can not be promoted to a higher level. In the book "cloud network integration: digital information infrastructure in the era of computing power", written by Li Zhengmao, general manager of China Telecom, etc., it summarizes the unprecedented changes in the paradigm of society, life and scientific research promoted by computing power.
In 1765, the appearance of Jenny spinning machine opened the prelude to the industrial revolution and officially opened a new era of mechanization instead of manual work. 20 years later, Watt invented the circular steam engine, which promoted the popularization and development of machines, and the factory has become an important organizational form of industrial production; In the 19th century, the invention of electric power marked the arrival of the second industrial revolution. The traditional industrial industries split into emerging markets such as aviation, automobile, petroleum and electrical appliances, and the industry took on a new look; After entering the 20th century, with the advent of the first electronic computer as the beginning, the third industrial revolution came quietly, smart phones, UAVs and chips appeared on the stage one after another, and the level of human informatization has been greatly improved. In 2013, Germany put forward the "industry 4.0" plan at the Hannover Industrial Expo, which is regarded as a landmark event of the fourth industrial revolution.
The fourth industrial revolution is a change in the production mode driven by digital technology represented by artificial intelligence, big data, cloud computing and the Internet of things, so as to realize the intelligent connection between factories and between factories and users, and make the production mode change from standardization to networking, informatization and customization.
When it comes to the fourth industrial revolution, we have to mention intelligent manufacturing. Intelligent manufacturing is the informatization and innovation of the manufacturing value chain and the deep integration of industrial technology and informatization, including the intellectualization of product development, application equipment and production line. Around intelligent manufacturing, emerging fields such as intelligent factory, intelligent management, intelligent decision-making and intelligent logistics have been derived. The wide application of network and computing power is the key guarantee to realize intelligent manufacturing.
In the field of industrial manufacturing, factories can monitor the production environment by installing temperature, humidity and other sensing equipment. In case of abnormality, they will give an alarm and adjust the environment. In the production process, industrial intelligent manufacturing can monitor the whole production process such as equipment status, production line status, product status, quality traceability, production scheduling and materials in real time through the camera, find the faults in each link in time, and record and store them for subsequent analysis and improvement.
Siemens Central Research Institute once demonstrated a part of double arm robot in Munich. With the help of high automation technology of artificial intelligence, it can work independently and cooperate without programming for product manufacturing. For the introduction of this robot, human engineers only need remote guidance through the network, and the repeated and fine work can be completed automatically by the intelligent robot, so as to reduce the probability of error and reduce the workload of manual operation.
With the development of the industry, there are more and more networking devices. Cameras, sensors and intelligent robots will be widely used in industrial manufacturing. As a derivative of the digital process, the massive data carrying the production value will continue to flow to the scattered computing nodes. With the support of algorithm, computing power, storage and network, the data processing system of manufacturing industry will become smarter and smarter.
The original weather forecast is a metaphysics, which is judged purely by people's life experience. Some popular proverbs can also reflect people's concern about the weather in the early days, such as "don't go out in the morning, the sunset goes thousands of miles", "clouds fly all over the sky, rain and snow keep falling". The birth of weather map indicates the beginning of modern weather forecast. At the same time, the emergence of telegram enables the observation data of each meteorological station to be transmitted and concentrated quickly, making it possible to draw real-time weather forecast. Weather charts are better at forecasting short-term weather, and there is nothing to do in the face of medium and long-term weather forecasting. Then the advent of numerical weather forecast solved this problem.
Numerical weather forecast is a method to predict the state of atmospheric motion and weather phenomena in a certain period of time in the future by solving the equations of hydrodynamics and thermodynamics describing the process of weather evolution through numerical calculation by large-scale computer under certain initial and boundary value conditions according to the actual situation of the atmosphere.
It is not easy to solve the weather equations. Richardson, a British mathematician, wrote about his calculation process in his book weather forecast using numerical methods. Richardson made his first attempt at numerical prediction between 1916 and 1918, during which he organized a large number of manpower and took 12 months to complete with the help of hand-held computers. From this, he came to the conclusion that if a person keeps calculating day and night, it will take 64000 days; Even if 64000 people are allowed to operate in parallel, it will take a whole day to estimate the weather in the next 24 hours.
Because the numerical prediction needs to process a large number of observation data, it must rely on the equipment with very high computational power. The birth of electronic computer perfectly solves this problem, and with the improvement of computing power, the time limit of numerical prediction increases by 1 day every 10 years. So far, numerical prediction has become the mainstream prediction model.
At the same time, with the demand for meteorological observation and the upgrading of observation equipment, the current observation range is expanded from the ground to thousands of meters high, and the observation elements are also expanded from temperature, humidity, wind speed and air pressure to more complex physical quantities. At the same time, due to the evolution of urbanization and the transformation of industry, the concentrations of carbon monoxide, nitrogen dioxide, sulfur dioxide and other gases in the atmosphere are also included in the monitoring range. These are inseparable from the upgrading of computing equipment.
In fact, news about the upgrading of high-performance computers in major meteorological institutions around the world is also common in the newspapers. At the beginning of 2020, the UK announced that it would spend £ 1.2 billion to build a special supercomputer for weather forecasting in the next decade to speed up the speed of weather forecasting; In the same year, the National Oceanic and Atmospheric Administration announced the purchase of two high-performance computers with 12pflops peak computing power for meteorological analysis and prediction; In China, many supercomputing centers cooperate with local meteorological departments to provide people with refined meteorological prediction services.
With the improvement of living standards, people put forward higher requirements for the temporal and spatial accuracy of weather forecasting, and the amount of data in the meteorological industry shows a large-scale growth trend. How to introduce artificial intelligence into weather forecasting and how to use artificial intelligence technology to improve forecasting accuracy have become a hot topic in the world.
Some studies have shown that relying on the weather prediction model of artificial intelligence can significantly improve the prediction level of catastrophic events such as typhoons, thunderstorms and hurricanes. Based on the experimental data over the past 20 years, the Japanese research team drew 1.05 million low-pressure cloud images, and then processed by the deep convolution neural network of artificial intelligence to obtain the resolution model of the global cloud system. Through this model, the occurrence of low-pressure in the Northwest Pacific hot belt in summer can be identified one week in advance. In China, there are also some applications about using artificial intelligence to realize minute level and street level weather forecasting.
With the maturity of technology and the improvement of computing power, more application scenarios and more meteorological service products will be produced in the future, which will bring more new opportunities to the industry and provide more valuable information for people's production and life.
Scientific research has different paradigms in different times. Breaking the old paradigm framework is the paradigm revolution. In 2007, computer scientist Jim gray mentioned four paradigms of scientific research in his speech, namely, the first paradigm focusing on describing natural phenomena, the second paradigm calculating key variables, the third paradigm verifying and predicting by computer, and the fourth paradigm relying on computer analysis to get conclusions. Others call the fourth paradigm scientific research based on big data. Its arrival indicates that the research method of "conclusion first simulation verification" we are used to is reconstructed, and the importance of high-performance computing power in scientific research is more prominent.
It also needs computers to participate in the scientific research process. The third paradigm emphasizes theory first, that is, put forward possible conclusions first, and then prove the rationality of assumptions through algorithms, data and simulation; The fourth paradigm focuses on data first. Based on a large amount of existing data, and then through the complex calculation process of high-performance computer, it obtains the previously unknown conclusion. It is a completely data-based scientific research model. If the human brain is the protagonist in the third paradigm, computers and data occupy the main position in the fourth paradigm.
The fourth paradigm subverts people's thinking conventions for thousands of years. The human brain prefers to analyze the causal relationship of things and associate certain phenomena with the underlying reasons; Computers are not good at causal logic, but they are better at exploring correlation. Victor Mayer Schoenberg mentioned in the big data era that the biggest change in the big data era is to give up the desire for causality and focus on correlation instead.
The fourth paradigm has spawned a series of revolutionary achievements in many fields.
Neutrino is a basic particle widely distributed in nature. It can pass through human body, walls, mountains, lakes and stars. It is difficult to capture and detect. It is called ghost particle. However, scientists have spent more than 20 years from predicting the existence of neutrinos to discovering them. Conventional physical experiments and space exploration methods can not observe and study neutrinos. Thanks to the establishment of supercomputing, in 2015, Chinese scientists used tianhe-2 supercomputing system to complete the numerical simulation of cosmic neutrinos and dark matter with 30 trillion particles, revealing the long evolution process of about 13.7 billion years since the big bang in 1600, and helping mankind open the door to explore the origin of cosmic rays.
In February 2020, MIT published a breakthrough research achievement in the top academic journal Cell, claiming that a new antibiotic was discovered by using artificial intelligence technology, which may be one of the most powerful antibiotics at present; The alphafold system developed by Google's deepmind team can predict the three-dimensional structure of proteins based on a large number of genomic data, providing an important basis for the research and development of biomedicine; The black hole image released in 2019 was released to the public after two years of data calibration and analysis by supercomputer, which successfully confirmed Einstein's general theory of relativity.
We can see that the fourth paradigm of scientific research, with the help of the development trend of increasingly perfect computer performance, has set off a revolution of the intersection and integration of scientific methods and information technology.Scientists can easily obtain a large amount of observation and experimental data from nature to society and from micro to macro. When these data can flow freely, be processed intelligently, be integrated and related, and when the analysis and processing means of data are available and supplied on demand like hydropower, the discovery and dissemination of knowledge becomes more frequent and meaningful.
In October 2021, Facebook CEO Zuckerberg announced at the Facebook connect conference that the company would be renamed meta, which is derived from the English word metaverse of metauniverse. As soon as the news came out, it attracted great attention from all walks of life in the industry.
From the literal meaning of metaverse, metaverse = meta + verse. Among them, meta corresponds to the "Yuan" of the meta universe, and the original meaning of this English prefix is "transcendence" and "change"; "Universe" and "universe" mean "universe". Metaverse can be understood as the universe beyond the universe and the universe above the real world. There are different opinions on the specific definition of the meta universe. A more representative definition is: the physical reality enhanced by virtual, which presents the characteristics of convergence and physical persistence, and the 3D virtual space based on the future Internet with the characteristics of connection perception and sharing.
Metauniverse has absorbed the achievements of information revolution (5g / 6g), Internet Revolution (Web3.0), artificial intelligence revolution, VR, AR and MR, especially the virtual reality technology revolution including game engine, showing the possibility of building a holographic digital world parallel to the traditional physical world to mankind; It has triggered the interaction of information science, quantum science, mathematics and life science, and changed the scientific paradigm; Promoted the breakthrough of traditional philosophy, sociology and even humanities system; It integrates blockchain technology, NFT and other digital financial achievements, enriching the transformation mode of digital economy.
Therefore, the supporting technology of the meta universe can be said to be a great achievement of digital technology. If the meta universe is compared to human beings, Internet of things technology, display interaction technology, three-dimensional modeling technology and real-time rendering technology are external expression technologies, while artificial intelligence technology, blockchain technology, network and computing technology are internal technologies that support the movement of the meta universe internally, In particular, network and computing technology can be said to be the internal skeleton supporting the meta universe.
In order to seamlessly switch between yuanuniverse and the real world through digital technology, it needs to transmit a huge amount of data. Taking VR equipment as an example, the screen effect should be eliminated when using. The image should reach at least 16K resolution and meet the refresh frequency above 120Hz. This means that at least 15GB of data is generated per second. The time delay of 5g cloud VR service model measured in the current network is up to 70ms, but if you want people not to feel dizzy when using XR, the end-to-end delay of immersive XR must be less than 20ms, which requires the further development of the network. In addition, the scale of interactive users that metauniverse needs to support will reach 100 million. At present, large-scale online games use client software, and take the game operator's server and user's computer as the operation mode of processing terminal. It has high requirements for the performance of computer terminal, forming the user's use threshold, which limits the user's access. At the same time, the terminal server has limited carrying capacity, which is difficult to support large-scale users online at the same time.
Only the ubiquitous network connection, low latency and large bandwidth communication quality can make people shuttle freely between the real and virtual world. The mature development of communication network can help users obtain rendering quality comparable to that of local host with ultra-low delay; Edge computing provides a scheme to support the huge content system and computing requirements of the virtual world and realize the smooth operation of the virtual world. 5g. The large-scale construction of Internet of things, the completion of large-scale data center and the promotion of "counting from east to west" have provided soil for the construction of meta cosmic infrastructure in terms of network data transmission and computing power. How to further release the resource potential and realize the flexible allocation and efficient coordination of technology and resources is the key problem to be solved by the meta universe infrastructure.
Some people believe that the second half of the Internet is the home of metauniverse, and the industrial chains such as online games, social Internet, cloud computing and intelligent wear will usher in unprecedented prosperity. With the joint improvement of network and computing power, the technical threshold of the meta universe will continue to be reduced, and the accessibility of the meta universe will be greatly improved. Under the strong influence of national policies and market demand, the popularity of public and private cloud in China has continued to rise in recent years, providing technical soil for the development of meta universe technology. In the future, there will be more possibilities in the digital world running on the ubiquitous, intelligent, agile and integrated computing architecture.
The rapid development of digital economy, the wide range of radiation and the deep degree of influence are unprecedented. Digital technology is deeply affecting the development process of human society. In particular, the digital information infrastructure based on computing power has increasingly become the base of scientific and technological progress and economic and social development. It has changed human production mode, life mode and scientific research paradigm. It represents a new development level of human wisdom. Its strategic position has been highly valued by the country and triggered a new round of competition among countries all over the world, Both inside and outside the industry are also competing to accelerate the deployment of computing facilities in order to seize the development opportunity.
This book mainly relies on the author's more than 30 years of working experience in the communication and information industries, starts from the perspective of telecom operators, and launches an all-round discussion on the era of computing power, computing power application, computing power system, computing power trend and digital information infrastructure under cloud network integration. This book is not only suitable for practitioners in the field of communication and information, but also for college teachers and students majoring in communication and information technology. It will also bring inspiration and help to people from all walks of life interested in digital economy and information infrastructure.