《1 Engineering research fronts》

1 Engineering research fronts

《1.1 Trends in top 10 engineering research fronts》

1.1 Trends in top 10 engineering research fronts

The top 10 engineering research fronts in the field of information and electronic engineering are summarized in Table 1.1.1, including the subfields of electronic science and technology, optical engineering and technology, instrument science and technology, information and communication engineering, computer science and technology, and control science and technology. Among them, “silicon-based integrated photonic devices for optical interconnects, optical computing, and optical sensing,” “5G large-scale antenna array wireless transmission theories and technologies,” and “artificial micro/nanostructures and their manipulation of the optics and electromagnetic fields” are among the popular topics published by Clarivate Analytics; “brain- inspired intelligence” is a product of the clustering analysis from the projects supported by the governments of eight countries or regions, including China, the USA, EU, Canada, and Russia, during the period 2016–2018; “new- generation neural networks and their applications” is a result of the analysis of the topics of prominence from 2015 to 2018 in the field of artificial intelligence (AI) in SciVal; the five other fronts are recommended by researchers.

The number of core papers published from 2013 to 2018 related to each front is listed in Table 1.1.2. Among them, “new-generation neural networks and their applications” is the most significant front based on the rapid increase in the number of core papers published in recent years.

(1)  Brain-inspired intelligence

AI has become the core driving force in the new round of technological revolution and industrial transformation. Varying from the current algorithm-centered AI, the brain- inspired intelligence attempts to imitate, learn, and surpass the perceptive and cognitive functions of the biological brain, which is one of the important technical approaches for achieving the ultimate goal of AI, the artificial general intelligence.

Generally, there are two technical approaches for achieving brain-inspired intelligence, top-down functional analogy and bottom-up structural imitation, which are both opposing

《Table 1.1.1》

Table 1.1.1 Top 10 engineering research fronts in information and electronic engineering

《Table 1.1.2》

Table 1.1.2 Annual number of core papers published for the top 10 engineering research fronts in information and electronic engineering

and inseparable. The functional analogy based on cognitive science refers to the cognitive mechanism of the brain in designing new AI models. However, because it is significantly difficult to comprehend the cognitive mechanism, it is difficult to determine when it will make a breakthrough. The structural imitation based on neuroscience attempts to construct devices similar to the biological nervous system by accurate simulation of biological neurons, synapses, and neural circuits, and then produces similar functions through stimulation training. It is expected to achieve a breakthrough in a few decades.

In recent years, with the rapid progress of brain observation, analytic technigues and instruments, neuroscience, and cognitive science, several countries have launched the “Brain Project.” If breakthrough can be achieved in the analysis of drosophila brain, the fine analysis of the human brain is expected to be completed within about 20 years. Neuromorphic devices for fine simulation of biological neurons and synapses have emerged at the same time. The first machine capable of fine simulation of the human brain is expected to be completed by 2022. The structural imitation and functional analogy techniques are expected to achieve rapid docking and interaction, which will significantly accelerate the development of brain-inspired intelligence.

(2) Space-terrestrial integrated network

The space-terrestrial integrated network is an infrastructure based on the terrestrial network, which extends the space network to provide information service for the activities of various users, such as space-, land-, and sea-based users. It is a network system that uses network technology to achieve the interconnection of the Internet, mobile communication networks, and space networks, and provides various types of network services. Considering the large scale, the three- dimensional multilayered topology, a high degree of heterogeneity, and a wide variety of services of the space- terrestrial integrated network, the challenge in networking is to design a network architecture with clear structure, simple functions, and easy and efficient implementation. The network can adapt to the rapid developments and changes in communication technology and can support new applications that are emerging endlessly.

At present, certain foreign countries are focusing on the research of satellite-terrestrial network in the field of space-terrestrial integrated network, emerging trends including modifying the mode of development from traditional high-orbit constellation to medium/low orbit constellation, establishing new investment and financing modes, market operation modes, and building new satellite manufacturing factories for mass production. The research on the space-terrestrial integrated network in China was initiated only a few years ago. Owing to the limitations in global station construction and availability of orbital resources, the research is facing considerable challenges in networking technology. It is essential to use the technological infrastructure of China in the aerospace and Internet industries to actively explore this huge market.

When combined with the development status of the space- terrestrial integrated network of China, the primary research directions are: 1) land, sea, air, and space integration network architecture design; 2) protocol research for large-scale, highly heterogeneous space network; 3) lightweight handover mechanism research for highly dynamic mobility; 4) research on multi-dimensional network resource joint management technology.

(3)  Brain imaging technologies

The brain is intricate and complicated. There are not only hundreds of billions of neurons, but also millions of connections between them. Till now, the core functions of the brain, such as mood and emotions, are still unsolved problems. It is the key to overcome major diseases of the nervous systems, which seriously endanger the physical and mental health of people. It will also provide an important basis for the development of brain-like computing systems and devices, overcoming the limitations of traditional computer architecture, and determining the future developmental directions of AI.

Since the invention of the microscope at the end of the 16th century, every breakthrough in microscopic imaging technology has provided a developmental milestone to life science research. In recent years, several brain imaging technologies have been developed; moreover, significant progress has been achieved in imaging resolution, imaging speed, imaging depth, and imaging field-of-view. The new brain imaging technology focusing on the multiscale characteristics of the brain loop will provide key guidance and support for the task of the National Brain Plan in analyzing the structure and function of the brain loop on multiple levels. The primary research directions include: development of a high-throughput three-dimensional structure and function imaging and new sample processing technologies, as well as new methods for image data processing and analysis for the application of cell resolution to different biological whole brain nerves; rapid quantitative analysis of metatypes, connections, and activities; development of in vivo high- resolution optical imaging, and other new technologies with a wide range and deep penetration depth to achieve high spatial-temporal resolution of nerve activities in conscious and free-moving animals; development of new technologies of ultrafine imaging, such as photoelectric correlation to achieve ultrafine analysis and quantitative characterization of subcellular structures, such as synapses. In the future, further improvement of the imaging depth of living brain imaging, achieving high-speed and high-resolution three-dimensional reconstruction of neural loop, and exploring accurate brain structure and function imaging are the expected development trends in brain imaging technology.

(4)  Synergetic sensing–communication–computation–control network: theory and methodologies

In future, the nodes of large-scale Internet of Things (IoT) have to be equipped with highly integrated sensing, communication, and computation capabilities. All the nodes participate in the processes of data acquisition, manipulation, transmission, and feedback, which are usually tightly coupled. This results in essential changes of the node model, traffic model, and control model, as well as the way of information evolvement in future networks, which further boosts the revolution of network architecture and finally results in a synergetic sensing–communication–computation–control (SSCCC) network. However, the theoretical foundation of such a network is rather weak and there is an urgent requirement for further research.

The primary research topics in this area include, but are not limited to, the following: 1) establishing new essential information metrics and key performance indicators that can satisfactorily reflect the performances of future large- scale networks; 2) developing the information-theoretic model and the analytical tools for the SSCCC networks; 3) investigating the rationale of information evolvement and the fundamental performance limits of the SSCCC networks with tightly coupled sensing, communication, computation, and control processes in addition to a varying large network state space under certain resource constraints. The goal is to reveal the origin of the information bottleneck in an SSCCC network and to achieve its network intelligence, moreover, to create a foundation for future large-scale IoT by developing the post-Shannon information theory.

(5)  Hybrid-augmented intelligence

The long-term goal of AI is to create machines that learn and think like human beings. Intelligent machines have become companions of human beings. The interaction between and hybrid of humans and machines are expected to become significantly common in our life. Owing to the high levels of uncertainty and vulnerability in human life and the open-ended nature of problems that humans are facing, no matter how intelligent machines are, they are unable to completely replace humans. Therefore, it is necessary to introduce human cognitive capabilities or human-like cognitive models into AI systems to develop a new form of AI, that is, hybrid- augmented intelligence. This form is a feasible and important development model for AI or machine intelligence.

Driven by the new round of science and technology revolution and grand changes in industry, AI has been rapidly used in industries, deeply involved with organizing and managing procedures, leading to disruptive changes to social org- anizations. Efficient collaborations between humans and intelligent systems for value co-creating and sharing are thought as the optimal pattern of AI being merged into the human society. Hybrid-augmented intelligence thus becomes the key supporting technology.

Hybrid-augmented intelligence is developing along two directions, human-in-the-loop hybrid-enhanced intelligence and brain and neuroscience inspired hybrid-augmented intelligence. These two directions share some common theoretical basis, while their research focuses are quite different. For human-in-the-loop hybrid-augmented intelligence, efforts are put on adopting human role in the computing pipeline of machine intelligence, to enhance the intelligence level of the machine. In contrast, the research on brain and neuroscience inspired hybrid-augmented intelligence focuses on how to build brain-inspired computational models and self-learning.

Currently, rapid progress has been made in hybrid-augmented intelligence. Human-centered AI, a good example of human- in-the-loop hybrid-augmented intelligence, has been wildly applied in finance, medical treatment, management, etc. Meanwhile, intersection of brain science, neuroscience, and AI has been mutually promoted. More and more new cognitive computing models and computing architectures have been proposed. Hybrid-augmented intelligence has become the main feature of the new generation AI and is playing the leading role.

(6)   Silicon-based integrated photonic devices for optical interconnects, optical computing, and optical sensing

Silicon photonics is a technology for developing and integrating photonic and optoelectronic devices based on silicon or using silicon as the substrate (e.g., SiGe/Si, silicon- on-insulator), using existing complementary metal-oxide semiconductor compatible processes, which can process and manipulate photons for effective optical interconnects, optical computing, and optical sensing. This technology shares the characteristics of integrated electronic circuit, such as ultra-large-scale integration and ultra-high-precision manufacturing, as well as the advantages of photonic technology, such as ultra-high-rate operation and ultralow power consumption. Among them, different kinds of silicon- based integrated photonic devices are the core foundation and premise of this technology, including silicon-based optical transmitters, optical detectors, optical modulators, and waveguide devices.

Silicon-based optical transmitters are used to generate light waves as information carriers, such as silicon-based light-emitting diodes and silicon-based lasers; moreover, the current research focuses on efficient light-emitting mechanisms based on silicon doping and defect regulation, and the integration and regulation of new light-emitting nanomaterials. Silicon-based optical detectors are used to receive optical signals and convert them to electrical signals, which are then transmitted to computing cells; further, the current research focuses on improving detection sensitivity, reducing dark currents, and increasing bandwidth. Silicon- based modulators are used to load electrical signals from computing units onto optical waves; further, the current research focuses on increasing the modulation rate, reducing insertion loss, and increasing the modulation depth. Silicon- based waveguide devices are the channels for optical transmission, including routers, wavelength-division multiplexers, and polarization multiplexers; the current research primarily focuses on designing and fabricating new structures to improve the capacity of information transmission.

(7) 5G large-scale antenna array wireless transmission theories and technologies

The large-scale antenna array (LSAA) based wireless transmission technology is one of the key 5G technologies. It is different from other existing wireless communication systems. In this system, the base station (BS) is equipped with an LSAA and provides services for multiple users using the same time-frequency resources. The number of antennas usually ranges from dozens to hundreds and is one to two orders of magnitude higher than that in existing wireless communication systems. Under the circumstances of multiple cells, time division multiplexing, and infinite antennas in every BS, the LSAA-based system has different characteristics when compared to other systems in a single cell and with finite antennas. While considering the antenna configuration, on one hand, these antennas can be collocated in a single BS to form a collocated LSAA-based system; on the other hand, these antennas can be distributed in multiple nodes to form a distributed LSAA-based system.

The LSAA-based wireless transmission technology has four advantages. First, the LSAA-based system has higher spatial resolution than the existing multiple-input multiple-output (MIMO) system, which implies that it can further explore resources in the space domain. Using the degree of spatial freedom from LSAA-based wireless transmission, more users can simultaneously communicate with the BS using the same time-frequency resources. Thus, the spectrum efficiency can be significantly improved without increasing the BS density. Second, the LSAA-based system can provide a narrower beam; thus, the co-channel interference can be significantly reduced. Third, its transmitted power can be reduced by a wide margin; consequently, the power efficiency is improved. Fourth, if the number of antennas is sufficiently large, the detection performance is optimal even when the simple linear precoder and detector are used. Additionally, noise and uncorrelated interference can be omitted.

To explore the potential advantages provided by the LSAA- based wireless transmission technology, a channel model that is suitable for practical 5G application scenarios was proposed. Its effect on the capacity was also analyzed. Under the circumstances of the practical channel model, appropriate pilot overhead, and acceptable implementation complexities, the spectrum efficiency and power efficiency were investigated; moreover, the optimal wireless transmission scheme, estimation methods of channel state information, and joint resource scheduling methods for multiuser spatial resource sharing were intensely investigated. There are 64 antennas in commercially deployed sub-6 GHz BSs. To further improve efficiency and coverage, the evolving 5G and 5G millimeter wave (mmWave) systems are expected to be equipped with more than 256 antennas.

(8)  Artificial micro/nanostructures and their manipulation of the optics and electromagnetic fields

Electromagnetic metamaterials are artificial materials created by homogenizing a series of subwavelength unit cells, whose local electromagnetic response can be manipulated. In infrared and visible frequencies, such metamaterials are composed of micro/nanostructures. The exotic properties of the artificial materials and their strong capability to efficiently control the electromagnetic waves have enabled a broad spectrum of applications, such as meta-hologram, metalens, and carpet cloaking. In the past ten years, the studies on metamaterials and micro/nanostructures for engineering of electromagnetic wave-based applications were selected twice among the annual top 10 scientific breakthroughs by Science.

The past decade has also witnessed continuous advancements in novel design and fabrication techniques of micro/nanostructures. This has become a growingly important scientific frontier and inter-disciplinary research area, with the strong connections of solid-state physics, materials science, mechanics, applied electromagnetics, optoelectronics, etc. As a key platform of electromagnetic wave technology, the artificial micro/ nanostructures is positively expected to lead the revolution of modern information technologies, secure national defense industry, emerging energy-harvesting technology, high- end semiconductor/chip nano-fabrications, etc. Currently, the policy makers, academia, and industries world-wide have paid remarkable attention to research, development, and fabrication of such artificial micro/nanostructures. For example, the USA Department of Defense has focused on artificial micro/nanostructures, under the branches of “Metamaterials and Plasmonics” and “Nanoscience and Nanoengineering,” from among their officially deemed six disruptive basic research areas. In Europe, more than 50 leading scientists and laboratories have received tremendous and highly selective funds for their related research. Even in Japan, despite the economic downturn and significant downscaling of research funds, at least two technical research projects on artificial micro/nanostructures have been generously funded. In the industry, top-notch semiconductor manufacturers, such as Intel, Advanced Micro Devices, and International Business Machines Corporation (IBM), have established a joint fund to support the research and development of artificial micro/nanostructures to control light and electromagnetic fields.

(9)   Quantized precise metering/measurement and related theory

Since May 20, 2019, the seven units of the International System of Units (SI) were redefined in terms of constants that describe the natural world. The units, second, meter, kilogram, ampere, kelvin, mole, and candela, are correspondingly defined by the following constants: unperturbed ground-state hyperfine transition frequency of the cesium 133 atom (ΔνCs), speed of light in vacuum (c), the Planck constant (h), elementary charge (e), the Boltzmann constant (k), the Avogadro constant (NA), and the luminous efficacy of monochromatic radiation of frequency 540 × 1012 Hz (Kcd). This indicates that an era of quantum metrology has originated. The new definitions will assure the future stability of SI and build a firm foundation for the use of new technologies, including quantum technologies.

The precise quantum measurement technologies are increasing based on the new SI definitions. These technologies employ quantum systems, quantum properties, or quantum phenomena to measure a physical quantity, which demon- strate the advantages of significantly high sensitivity and precision. A different class of applications has emerged with quantum measurements for various physical quantities ranging from magnetic and electric fields to time, frequency, and temperature. In the future, with the development of the quantum theory and technology, the quantum metrology/ measurement is expected to advance toward metrology democratization, considerably low noise, and significantly high accuracy. This technology will gradually change the traditional measurement capabilities, enabling higher sensitivity and precision, including atomic observations up to macroscopic length scales.

(10) New-generation neural networks and their applications

Inspired by brain research in biology and neuroscience, artificial neural networks use nonlinear mapping functions to model the input-output transformation process of neurons and grow into a category of machine learning methods. The rapid development of new-generation neural networks is attributed to the breakthrough of deep learning based neural networks. Studies on deep neural networks primarily include learning theory of neural networks (such as generalization ability and regularization methods), supervised/unsupervised deep neural networks, convolution neural networks, sequential neural networks, attention model, compression and acceleration of neural networks, neural architecture search, and new models of neural networks (such as memory networks, generative adversarial networks, and deep reinforcement neural networks). Deep neural networks have been successfully applied in various fields of AI, such as representation learning, computer vision, pattern recognition, speech recognition and synthesis, natural language processing, and robotics. Owing to its significant performance improvement in image classification, speech recognition, and machine translation, deep neural networks are widely deployed in different industrial applications. The challenges of current research on neural networks include certain basic defects, such as the lack of learning autonomy, high cost of the training process, poor adaptability to the open dynamic environment, and the low level of privacy protection. The efforts that boost the research of neural networks may involve: 1) developing new models of neural networks based on the research on brain cognition; 2) interpretability of deep neural networks; 3) small sample size and meta-learning for deep neural networks; 4) adversarial game and security of neural networks.

《1.2 Interpretations for three key engineering research fronts》

1.2 Interpretations for three key engineering research fronts

1.2.1 Brain-inspired intelligence

The research and development of brain-inspired intelligence can be traced back to a series of brain-based devices proposed and developed by the Nobel Prize winner, Gerald Edman, in the 1980s, and to Neuromorphic Engineering initiated by Professor Carver Mead at the California Institute of Technology. Since the start of the new century, developed countries have launched the development of neuromorphic computing systems. On October 29–30, 2015, the US Department of Energy convened a forum for experts on the topic of “Neuromorphic Computing: From Materials to System Architecture.” In 2016, three large-scale neuromorphological computing systems were implemented one by one: the BrainScaleS system of the Heidelberg University in Germany, the SpiNNaker system of the University of Manchester, and the TrueNorth based chip system from IBM. There are other neuromorphic systems, including the Neurogrid of Stanford University, Si elegans of the University of Ulster, UK, and the neuromorphic chip and Loihi-based systems developed by Intel in recent years. According to the article A

Survey of Neuromorphic Computing and Neural Networks in Hardware published in May 2017, the number of papers on neuromorphology technology has increased rapidly since 1985 with a total of 2682 papers, indicating that the implementation of brain-like systems technology is the predominant force in the development of brain-inspired intelligence.

The related research on brain-inspired intelligence in China was conducted over a period of ten years. In September 2015, Beijing launched the “Brain Science Research” project, for two major tasks, “brain cognition and brain-like computing,” which considered nine tasks at three levels, i.e., theoretical basic research, brain-like computer development, and brain- like intelligence application, including the brain structure analysis platform, cognitive function simulation platform, neuromorphic devices, brain-like processors, machine learning chips, brain-like computers, audiovisual perception, autonomous learning, and natural conversation. Researchers in Beijing worked together to challenge the major common technologies and have achieved important results. For example, Shi Luping and his colleagues at Tsinghua University proposed a paradigm framework for brain-like hybrid computing and developed a brain-like chip called “Tianjic.” The results were published as the cover paper of Nature in 2019. Huang Tiejun and his team at Peking University proposed a bionic visual spike coding model for simulating the mechanism of the retina; moreover, in 2018, they developed a full-time and ultra-fast retina-like chip that was thousands of times faster than the human eye. At the national level, the implementation plan of the major scientific and technological projects of “brain science and brain-like intelligence” presented since 2016, was formally compiled and is expected to start soon. In 2018, the Ministry of Science and Technology of China issued a national plan, i.e., Science and Technology Innovation 2030—New-Generation AI, which clearly considers neuromorphic technology and chips as an important research direction.

Neuroscience research in the USA has a strong foundation. Brain-inspired intelligence research is also conducted in academic institutions and enterprises. The numbers of core papers and citations in the USA account for more than half of the global total. UK accounts for nearly 20% of the core papers globally. Germany, Canada, and the Netherlands each account for about 10%, and China, Sweden, Spain, and Italy each account for about 5% (Table 1.2.1). The international cooperation targets of China are primarily the USA and Canada. The cooperation among other countries is the same (Figure 1.2.1). The production institutions of core papers are relatively centralized (Table 1.2.2 and Figure 1.2.2). Among the institutions which published no less than 10 core papers, beside the Swedish Karolinska College, a major brain science center, all others are well-known universities in the field of brain science and AI, including Harvard University, Yale University, McGill University, Oxford University, Stanford University, Edinburgh University, Cambridge University, and Toronto University (Table 1.2.2). The number of citing papers in the USA accounts for more than one-third, UK and China both account for more than 10%, and the distribution of other countries is basically the same as that of the producing countries, indicating that China is obviously on par with the top countries in the field of brain-inspired intelligence (Table 1.2.3).

《Table 1.2.1》

Table 1.2.1 Countries or regions with the greatest output of core papers on “brain-inspired intelligence”

《Table 1.2.2》

Table 1.2.2 Institutions with the greatest output of core papers on “brain-inspired intelligence”

《Figure 1.2.1》

Figure 1.2.1 Collaboration network among major countries  or regions in the engineering research front of “brain-inspired intelligence”

《Figure 1.2.2》

Figure 1.2.2 Collaboration network among major institutions in the engineering research front of “brain-inspired intelligence”

《Table 1.2.3》

Table 1.2.3 Countries or regions with the greatest output of citing papers on “brain-inspired intelligence”

Among the top 10 institutions with the greatest output of citing papers, six come from the USA (Table 1.2.4).

1.2.2 Space-terrestrial integrated network

Considering the large scale of the space-terrestrial integrated network, three-dimensional multi-layered topology, high degree of heterogeneity, and the wide variety of services, networking focuses on designing a network architecture with a clear structure, simple functions, and easy and efficient implementation. This is the primary problem that the space- terrestrial integrated network is required to solve.

At present, the development priorities of major countries and regions in the world are as follows: (1) The USA is committed to the large-scale construction of a commercial integrated network, such as the large-scale manufacturing and launch of low-cost low-orbit satellites by Starlink, and the promotion of the Google Loon project for commercialization. (2) The EU is committed to the integration of the satellite-terrestrial network and the 5G network architecture, focusing on the combination of software defined networking (SDN)/network functions virtualization (NFV). There are several projects under the Horizon 2020 plan that have launched demos. (3) China has established a major project on space-terrestrial integrated information network and a low-orbital satellite network construction plan.

At present, research on the integration technology of the space-terrestrial network is conducted primarily in the following directions:

(1)  Integrated network architecture design for land, sea, air, and space

The infrastructure of the space-terrestrial integrated network is divided primarily into high-, medium-, and low-orbit satellites, high-altitude aircraft, marine mobile equipment, and ground equipment. The advantages and disadvantages of different facilities in terms of coverage, transmission delay, bandwidth cost, capacity, frequency, etc. are different. To efficiently carry various types of service, determining the characteristics of various types of equipment to form a composite collaborative network and fully utilizing the new technologies in the network field, such as SDN, to improve system controllability, are key issues in the design of the space-terrestrial integrated network architecture.

(2)   Protocol research for large-scale, highly heterogeneous space network

At present, various network protocols, such as the Consultative Committee for Space Data Systems (CCSDS), Delay Tolerant Network (DTN), Snapshot, and Internet Protocol (IP), have been proposed in the field of space-terrestrial integration. A spacecraft can adopt the CCSDS protocol. In the case of intermittent connection, the DTN protocol can be used. In the case of regular motion, the Snapshot protocol can be used. The terrestrial users can use the IP protocol to support broadband networking applications. Recently, researchers also proposed the use of a content centric networking (CCN) in this area. This protocol family requires further enrichment and adaptation for the specific scenarios of the space-terrestrial integrated network, which is a key challenge that requires further research.

《Table 1.2.4》

Table 1.2.4 Institutions with the greatest output of citing papers on “brain-inspired intelligence”

(3)   Lightweight handover mechanism research for highly dynamic mobility

Considering the characteristics of the constructed network nodes, especially low-orbit satellites with highly dynamic motion, user-oriented multi-satellite, and multibeam frequent switching, it is necessary to focus on highly dynamic motion and light-weight mobility handover mechanisms to effectively reduce the delays and data loss during the handover between networks, which can improve network mobility performance when space network resources are limited.

(4)   Research on multi-dimensional network resource joint management technology

For the integration of space-based networks and terrestrial networks, technologies such as SDN/NFV have to be applied to the integrated network to achieve the key technologies of multidimensional network resource virtualization slicing and service quality assurance, application-driven network control, on-demand network resource scheduling, and safe and reliable network management, which can achieve efficient control and interconnection of satellite Internet and terrestrial Internet.

The top three countries with the greatest output of core papers on the research front of “space-terrestrial integrated network” are the USA, China, and Germany (Table 1.2.5). According to the main production institutions of the core papers (Table 1.2.6), the top three institutions are Southeast University, California Institute of Technology, and Tsinghua University. From the cooperation network of major countries or regions (Figure 1.2.3), it can be observed that all the concerned countries demonstrate close cooperation. From the cooperation network between the main institutions (Figure 1.2.4), it can be observed that the major cooperation is between Southeast University and the PLA University of Science and Technology. From the statistical results of the primary output countries and regions that cite the core papers (Table 1.2.7), China, the USA, and Germany were ranked the top three. Among them, China was ranked first with 1753 papers, accounting for 26.93%. From the list of primary institutions that cite core papers (Table 1.2.8), the top three institutions are Chinese Academy of Sciences, the National Aeronautics and Space Administration, and Beijing University of Posts and Telecommunications.

1.2.3 Brain imaging technologies

In the second decade of the 21st century, we are witnessing and experiencing a revolutionary change in the concept of brain intelligence. Because of the significant value of brain science research in the scientific, economic, social, and military fields, each developed country is attempting to become the global strategic command center for brain and cognitive technology. In recent years, the USA, the EU, and Japan have each successively issued a “brain plan,” expecting a major breakthrough in brain science, thereby providing a basis for the in-depth development of AI in the future and promoting the innovative development of emerging industries based on the integration of brain-like intelligence and the brain computer.

《Table 1.2.5》

Table 1.2.5 Countries or regions with the greatest output of core papers on “space-terrestrial integrated network”

《Table 1.2.6》

Table 1.2.6 Institutions with the greatest output of core papers on “space-terrestrial integrated network”

CALTECH: California Institute of Technology; NASA: National Aeronautics and Space Administration; ETH: Swiss Federal Institute of Technology Zurich

《Figure 1.2.3》

Figure 1.2.3 Collaboration network among major countries or regions in the engineering research front of “space-terrestrial integrated network”

《Figure 1.2.4》

Figure 1.2.4 Collaboration network among major institutions in the engineering research front of “space-terrestrial integrated network”

《Table 1.2.7》

Table 1.2.7 Countries or regions with the greatest output of citing papers on “space-terrestrial integrated network”

《Table 1.2.8》

Table 1.2.8 Institutions with the greatest output of citing papers on “space-terrestrial integrated network”

Brain science research has the dual characteristics of scientific frontier and comprehensive intersection. Brain imaging technology is an effective way to deeply analyze the brain functional connection group. Essentially, deep analysis of the brain functional connection group is a reverse engineering interpretation of the brain working principle. On this basis, this research is expected to develop a new computing system based on the brain structure and circuit principle, break the technical bottleneck of modern computers and AI while dealing with complex problems, and achieve self-organization and self-sufficiency, deep learning, and even a new type of neural AI system.

In recent years, the basic research of brain science has developed rapidly; moreover, the technology of AI and brain–computer interface is changing with each passing day. There is no doubt that brain science research has entered a golden age. With the development of brain science research, scientists have proposed higher requirements for brain imaging technology. They plan to focus on methods to integrate the macro, meso, and micro brain tissue structures, draw the brain function linkage map, seek the systematic grasp of brain tissue structure and function, and develop and optimize non-invasive tools, such as light, sound, electricity, and magnetogenetics, for treatmentment of nervous and mental diseases. The primary development directions of brain imaging technology are as follows: (1) high-resolution brain structure analysis methods and technologies, including the development of high-throughput three-dimensional structure and function imaging and new sample processing technologies, as well as new image data processing and analysis methods, to achieve rapid and quantitative analysis of the types, connections, and activities of different species of brain neurons with cell resolution; (2) large-scale in vivo high- resolution optical imaging with deep penetration and other new technologies that can achieve high-resolution analysis of nerve activities in conscious and free animals in space/ time; (3) photoelectric correlation and other new technologies that can achieve ultrastructural analysis and quantitative characterization of subcellular structures, such as synapses.

It can be observed from the countries or regions that primarily publish core papers in the area of “brain imaging technology” (Table 1.2.9) that the top three countries are the USA, UK, and Germany. According to the primary institutions that publish core papers (Table 1.2.10), the top three institutions are Harvard University, King’s College London, and Stanford University. It can be observed from the cooperation network among the major countries or regions (Figure 1.2.5) and among the major institutions (Figure 1.2.6) that a close corporation is demonstrated. According to the statistical results of the major output countries or regions (Table 1.2.11), the USA, China, and UK are ranked the top three. Among them, China is ranked second with 6423 papers, accounting for 11.77% of the total. According to the main output institutions of the citing papers (Table 1.2.12), the top three institutions are Harvard Medical School, University College London, and University of Toronto.

《Table 1.2.9》

Table 1.2.9 Countries or regions with the greatest output of core papers on “brain imaging technologies”

《Table 1.2.10》

Table 1.2.10 Institutions with the greatest output of core papers on “brain imaging technologies”

《Figure 1.2.5》

Figure 1.2.5 Collaboration network among major countries  or regions in the engineering research front of “brain imaging technologies”

《Figure 1.2.6》

Figure 1.2.6 Collaboration network among major institutions in the engineering research front of “brain imaging technologies”

《Table 1.2.11》

Table 1.2.11 Countries or regions with the greatest output of citing papers on “brain imaging technologies”

《Table 1.2.12》

Table 1.2.12 Institutions with the greatest output of citing papers on “brain imaging technologies”

《2 Engineering development fronts》

2 Engineering development fronts

《2.1 Trends in top 10 engineering development fronts》

2.1 Trends in top 10 engineering development fronts

The top 10 engineering development fronts in the information and electronic engineering field are summarized in Table 2.1.1, including the subfields of electronic science and technology, optical engineering and technology, instrument science and technology, information and communication engineering, computer science and technology, and control science. Among these 10 fronts, “systems and technologies for analysis and identification of images and videos,” “development of sensors based on the micro/nanoelectronic technology,” and “surgical robot technology” are published based on the analysis of Derwent Innovation of Clarivate Analytics; the seven other fronts are recommended by researchers.

The annual disclosure of core patents involved in the top 10 development fronts from 2013 to 2018 is listed in Table 2.1.2.

(1)    Millimeter wave (mmWave) ultra-high throughput technology

The mmWave ultra-high throughput technology achieves high-speed data transmission using mmWave frequency spectrum. The frequency range of the mmWave frequency band is 26.5–300 GHz, corresponding to 1–10 wavelengths. Using the spectrum resources within the above-mentioned bands, the data throughput can reach gigabits per second or even terabytes per second. It is universally acknowledged that a degradation of the performance occurs in microwave/ mmWave/terahertz integrated circuits (ICs); moreover, the cost

《Table 2.1.1》

Table 2.1.1 Top 10 engineering development fronts in information and electronic engineering

《Table 2.1.2》

Table 2.1.2 Annual number of core patents published for the top 10 engineering development fronts in information and electronic engineering