5G has been fully commercialized. With the continuous penetration of 5G in vertical industries, people's vision for 6G is gradually put on the agenda. Facing 2030+, 6G will fully support the digitization of the entire world on the basis of 5G, and combine with the development of artificial intelligence and other technologies to realize the ubiquitous and desirable wisdom, comprehensively empower everything, and promote the society to move towards a "digital twin" that combines virtuality and reality. "World, realize the beautiful vision of "digital twin, ubiquitous wisdom".

      Around this overall vision, the 6G network will spawn new application scenarios in three aspects: intelligent life, intelligent production, and intelligent society, such as twin digital human, holographic interaction, super transportation, synaesthesia interconnection, intelligent interaction, etc.

      These scenarios will require terabit-level peak rates, sub-millisecond-level latency experience, moving speeds exceeding 1,000km/h, and new network capabilities such as security endogenous, intelligent endogenous, and digital twins. In order to meet the higher requirements of new scenarios and new services, 6G air interface technology and architecture need corresponding changes.

                                                                                             01 Future Network Technology

       At present, with the in-depth integration of information and communication technology with big data and artificial intelligence, the further expansion of network ubiquity, the continuous improvement of user experience and personalized service requirements, and the continuous emergence of many new enabling technologies, the future network also presents Some major features and development trends are as follows.

1. Full spectrum communication

       With the continuous improvement of communication requirements, mobile communication networks need more spectrum. Since the spectrum below 6GHz has been allocated, the millimeter-wave frequency bands of 26GHz and 39GHz have also been allocated for 5G use. It is necessary to study higher frequency bands, such as THz and Visible light to meet the needs of higher capacity and ultra-high experience rate.

       Visible light usually refers to electromagnetic waves in the frequency band 430~790THz (wavelength is 380~750nm), and there are about 400THz candidate spectrum. Terahertz refers to electromagnetic waves in the frequency band 0.1~10THz (wavelength is 30~3000 microns), which has about 10THz candidate spectrum. Both of them have the characteristics of large bandwidth and are easy to realize ultra-high-speed communication, which is a potential supplement to the future mobile communication system.

▲Frequency distribution

       Both visible light and terahertz have large space transmission losses, so they are not suitable for long-distance transmission in terrestrial communications, but are suitable for providing larger capacity and higher rates in local and short-distance scenarios.

       In order to improve coverage, visible light communication can take advantage of its low power consumption, low cost, easy deployment and other characteristics, and combine it with lighting functions to achieve wider coverage by ultra-dense deployment; while terahertz communication has a short wavelength and a small antenna element size. The transmit power is low, so it is more suitable to be used in combination with ultra-large-scale antennas to form a terahertz beam with a narrower width and better directivity, which can effectively suppress interference and improve the coverage distance.

       From the perspective of the deployment of the entire 6G mobile communication network, it is necessary to comprehensively consider cost, demand and service experience, and effectively use all available frequency resources in different scenarios. Bands below 6GHz will still play an important role, especially providing seamless network coverage, etc. Millimeter waves will play a more important role, and THz and visible light bands will provide greater capacity and higher rate.

       Therefore, after visible light and terahertz communication are introduced into mobile communication networks, it is necessary to consider the deep integration networking of all frequency bands below 6 GHz, millimeter wave, terahertz, and visible light, so as to realize the dynamic complementarity of each frequency band, so as to optimize the overall service quality of the whole network and reduce the network power consumption.

2. The integration of space, sky and earth

       In the future, while greatly improving the user experience rate, the network must also meet the network service requirements of aircraft, ships, and other airborne and ship-borne Internet, ensure the service continuity of high-speed moving ground vehicles, high-speed rail and other terminals, and support immediate emergency rescue and disaster relief, environmental protection. The deployment of massive IoT devices such as monitoring, forest fire prevention, inspection in unmanned areas, and ocean-going container information tracking can meet the needs of low-cost coverage in sparsely populated areas. Therefore, the main form in the future is to expand the network coverage to a three-dimensional coverage network in natural spaces such as space, deep mountains, deep sea, and land. Therefore, it is necessary to build an air-space-ground integrated network to realize the three-dimensional "ubiquitous coverage" of the global communication network.

        The air-space-ground integrated network mainly includes three parts: the space base composed of satellites in different orbits, the space base composed of various aerial vehicles, and the ground base composed of satellite ground stations and traditional ground networks. It has wide coverage, flexible deployment, and ultra-low power consumption. , ultra-high precision and not easily affected by ground disasters.

▲Air-space-ground integrated network

      The 6G-oriented air-space-ground integration takes the satellite communication network as an important supplement and extension of the terrestrial communication network, and deeply integrates the two to significantly improve the user's air interface access capability and three-dimensional coverage capability. Through the satellite-ground resource coordination scheduling and satellite-ground seamless roaming of the air-space-ground integrated network, it can provide users with non-perceptual and consistent services, ensuring network resilience and robustness and green resource intensiveness.

3. DOICT integration

        6G is a new generation of mobile communication system that deeply integrates communication technology, information technology, big data technology, AI technology, and control technology, showing strong interdisciplinary and interdisciplinary development characteristics. The 6G vision of "digital twin, ubiquitous wisdom" requires end-to-end design from information collection, information transmission, information calculation, and information application. DOICT convergence will be the development trend of 6G end-to-end information processing and service architecture.

        The deep integration of ICT promotes the full-dimensional definability of the network, which is the foundation of the flexible network. The deep integration of DICT promotes the full penetration of artificial intelligence and big data into the network, which is the foundation of the intelligent network. The deep integration of DOICT promotes the development of deterministic networks and is the basis for automated systems and digital twin systems.

       DOICT will realize the deep integration of cloud, network, edge, terminal and industry on the basis of big data flow, create a credible environment represented by blockchain, improve the efficiency of resource utilization of all parties, and collaboratively upgrade cloud-edge computing capabilities, network capabilities, terminal capabilities and business capabilities.

4. The network can be reconfigured

        With the rapid development of mobile communication technology, business needs and scenarios are more diversified and personalized, and the future 6G network will adopt a more flexible and reconfigurable architecture design.

          On the one hand, based on shared hardware resources, the network allocates corresponding network and air interface resources for different services of different users to achieve end-to-end on-demand services. While providing ultimate services, it also realizes resource sharing to maximize resource utilization and reduce Network construction cost; on the other hand, the minimalist network architecture and flexible and scalable network features provide great convenience for subsequent network maintenance, upgrade and optimization, further reducing network operation costs for operators. In addition, facing the endogenous feature requirements of 6G intelligence, it also puts forward stronger computing power and scalability for the network.

5. Perception-communication-computing integration

        The integration of perception-communication-computing refers to an end-to-end information processing technology framework that synchronously executes information collection and information calculation in the process of information transmission, which will break the chimney-like structure of terminal information collection, network information transmission, and cloud-side computing. The information service framework is the technical requirement to provide highly coupled services of perception communication computing such as unmanned, immersive and digital twin.

        Perception-communication-computing integration is divided into two levels: function collaboration and function fusion. In the functional collaboration framework, perception information can enhance communication capabilities, communication can expand perception dimension and depth, computing can perform multi-dimensional data fusion and big data analysis, perception can enhance the performance of computing models and algorithms, communication can bring ubiquitous computing, computing Ultra-large-scale communication can be achieved.

        In the function fusion framework, the sensing signal and the communication signal can be integrated with waveform design and detection, and share a set of hardware equipment. At present, the integration of radar communication technology has become a hot spot, and the integration of terahertz detection capabilities and communication capabilities, as well as the integration of visible light imaging and communication, have become the potential technology trends of 6G. Perception and computing are integrated into a computing power-aware network, and computing and network integration realizes the network end-to-end definable and micro-service architecture.

         In the future, perceptual communication computing can realize functional reconfiguration based on the development of software-defined chip technology.

▲Perception-communication-computing integration application scenarios

       The application scenarios of perception-communication-computing integration include unmanned business, immersion business and digital twin business. In the field of unmanned business, it provides the ability of intelligent body interaction and collaborative machine learning; in the field of immersion business, it provides the ability of perception and rendering of interactive XR, and the ability of perception, modeling and display of holographic communication; in the field of digital twin business Provide perception, modeling, reasoning and control capabilities of the physical world, and provide personnel monitoring, human parameter perception and intervention capabilities in the field of body area networks.

02Wireless enabling technology

       In the face of new indicator requirements brought by new application scenarios, such as the peak rate of Tbps level, the user experience rate of Gbps level, and the delay of near-wired connections, it is difficult to meet the requirements only by relying on the existing 5G technology. The industry is also actively researching some new technologies, new architectures, and new designs, expecting to form some new breakthroughs. This chapter will analyze the potential key technologies of future wireless access networks from three aspects: basic transmission technology, protocol and architecture design, and autonomous network technology.

        As we all know, larger bandwidth can improve the peak rate of the system, but the improvement of spectral efficiency also depends on the development of physical layer transmission technology.

1. Distributed Super Massive MIMO

       After the introduction of ultra-massive MIMO, 4G/5G network capacity has been greatly improved, but due to path loss and inter-cell interference, the user experience at the cell edge still needs to be improved. Distributed ultra-massive MIMO extends the traditional centralized deployment method to distributed deployment, and introduces intelligent cooperation among multiple distributed nodes to realize joint scheduling of resources and joint transmission of data, as shown in the following figure. Through distributed deployment and intelligent collaboration, on the one hand, interference is effectively eliminated and signal reception quality is enhanced; on the other hand, coverage is effectively enhanced, bringing users a borderless performance experience. In the future, 6G networks will show great application potential, especially in higher frequency bands and dense deployment scenarios.

       The industry has theoretically demonstrated the advantages of distributed MIMO in improving channel capacity. Theoretical analysis shows that under the conditions of the same total number of antennas, total transmit power and coverage, distributed MIMO systems always have distributed nodes that are closer to users, and at the same time use the intelligent cooperation of scheduling and shaping, their performance is better than that of centralized ones. MIMO is more uniform, especially for edge users, the performance gain is more significant.

       Due to the significant increase in the antenna scale and the number of nodes, distributed ultra-large MIMO poses challenges to the ability of information exchange between nodes, joint cooperative node selection and shaping scheme design, algorithm complexity, and interference processing. The consistency of the transceiver channels between nodes also puts forward higher requirements, and further research on the air interface calibration scheme is required.

2. Smart metasurfaces

       Reconfigurable Intelligent Surface (RIS) controls electromagnetic waves through the structural units on the surface. By adjusting the parameters and positions of each structural unit, it realizes the adjustment of any electromagnetic wave reflection/shot-by-shot amplitude and phase distribution. It has positive significance in solving traditional wireless communication pain points such as non-line-of-sight transmission and reducing coverage holes.

       The figure below shows a schematic diagram of a wireless communication system assisted by RIS. The base station controls the RIS, and the RIS adjusts the amplitude and phase of its own structural unit based on the control, so as to realize the controlled reflection of the signal transmitted by the base station. Compared with traditional relay communication, RIS can work in full-duplex mode with higher spectrum utilization. RIS does not require RF links, does not require large-scale power supply, and will have advantages in power consumption and deployment costs.

                                                                                           ▲RIS auxiliary communication system

       The practical application effect of RIS in wireless mobile communication depends on the research maturity of metamaterials and the precision and efficiency of digitally controlled metamaterials. At the same time, the difficult estimation of the metasurface channel caused by passive characteristics, the practical joint precoding scheme of the base station and RIS, and the RIS network architecture and control scheme all need to be further studied.

3. Super Nyquist transmission technology

       In traditional communication systems, in order to avoid Inter-Symbol Interference (ISI, Inter-Symbol Interference), the Nyquist criterion is usually used, thereby limiting the rate of transmitted symbols. Ultra-Nyquist transmission technique error smell to find reference source. Send symbols at a faster rate, artificially introduce ISI during transmission, and then use a more advanced receiver to eliminate ISI through oversampling at the receiver, as shown in the figure below, thereby improving the actual transmission rate and spectrum utilization of the link.

▲The block diagram of the transmission and reception of the super-Nyquist transmission system

       The power spectral density of the super-Nyquist transmission signal is only related to the frequency response function of the transmit filter and does not expand the bandwidth. The following figure compares the bandwidth of the super-Nyquist transmission system and the traditional Nyquist system, in which the baseband time domain waveform is a rectangular wave, and the number of overlapping layers of the super-Nyquist transmission system is 4. As can be seen from the figure, the super-Nyquist transmission system does not change the distribution shape of the spectrum, that is, does not expand the bandwidth.

▲Bandwidth comparison of super-Nyquist system and Nyquist system

       In a multi-antenna antenna system, super-Nyquist transmission technology is used to generate delay between transmit antennas, and oversampling is used to create a virtual receive antenna, which can improve spatial multiplexing and diversity gain when the number of antennas on the user side is limited. Therefore, even single-antenna users can achieve spatial multiplexing gain. It can be seen from the figure below that when the signal-to-noise ratio is high, the virtual antenna system based on super-Nyquist transmission has obvious gain compared to the traditional MISO. When the signal-to-noise ratio is 10dB, a capacity gain of over 40% can be obtained.

▲Comparison of the capacity of super Nyquist transmission and traditional Nyquist transmission system

      The optimal decoding algorithm of super-Nyquist transmission technology is the Viterbi decoding algorithm based on maximum likelihood sequence estimation, but its complexity increases exponentially with the increase of overlapping degree. Therefore, low-complexity receiver design is crucial to the practical development of this system. At the same time, multi-carrier and large-scale antennas are still the mainstream technologies in the future. How to combine with OFDM/MIMO technology and consider the impact of actual multipath fading channels on the system needs to be discussed in depth.

     4. Transform domain waveform

      Waveform technology has played an important role in the air interface design of wireless communication systems of all generations. The performance of the OFDM waveform used by 4G and 5G systems depends on the orthogonality between its subcarriers. If the orthogonality between subcarriers is damaged by factors such as Doppler frequency offset, the performance tends to degrade significantly.

▲Schematic diagram of transform domain waveform principle

      Transform domain waveforms are designed to overcome the above-mentioned disadvantages of OFDM waveforms. Different from the traditional waveform scheme that the transmitted symbol is located in the classical time-frequency domain, the transform domain waveform considers that the transmitted symbol is located in other dual domains (such as delay-frequency, time-varying-Doppler, etc. dual domains), as shown in the following figure. Through the transformation between the dual domains, the transform domain symbols can achieve a multi-dimensional diversity effect, so that the unfavorable factors such as Doppler frequency offset in the OFDM waveform can be effectively utilized as a diversity degree of freedom to improve the transmission performance.
                                                                                ▲Comparison of transform domain waveform and OFDM performance

        The figure above shows the block error rate performance comparison between the down-transform domain waveform and OFDM under the assumption of ideal channel estimation in a 500km/h mobile environment. The CDL channel model is considered in the simulation, the sub-carrier spacing is 60 kHz, the channel coding is a convolutional code of 1/3 code rate, the number of sub-carriers is 128, and the transform domain waveform considers the joint processing of six consecutive time-domain OFDM symbols. The results show that the transform domain waveform can effectively deal with the Doppler frequency offset in the high-speed mobile environment and achieve better block error rate performance.

       Although related studies have shown that the transform-domain waveform scheme can achieve significant gains compared with the traditional OFDM-based waveform scheme in high-speed mobile scenarios, how to accurately restore the transmitted signal at a lower cost is an important topic in transform-domain waveform research. In addition, how to design efficient reference signals to accurately acquire multi-antenna channels with low overhead requires further research.

5. AI-driven physical links

       Since 5G communication, the intelligence of wireless network has become an important topic, aiming to realize more efficient allocation and utilization of network resources. As one of the main enabling technologies of current wireless network intelligence, AI technology is penetrating into the core network, network management, and physical layer and high-level protocol stack of the access network. Among them, physical layer AI generally refers to technical solutions that use artificial intelligence/machine learning methods to realize or enhance the functions of the physical layer of wireless networks.

       AI can be mainly applied to CSI processing, receiver design, and end-to-end link design at the physical layer. For example, neural networks in deep learning are used to learn compressed representations of high-dimensional CSI in wireless communication, thereby reducing the CSI feedback overhead; artificial neural networks are used to learn the inverse mapping from the received interference signal to the original signal, which can eliminate the need for explicit Channel estimation and equalization; jointly optimizing the transmitter and receiver in a specific channel environment can learn non-ideal effects in the channel and improve transmission performance.

       However, replacing traditional physical layer modules with AI modules in a "black box" manner will hardly surpass traditional designs in performance. In contrast, the idea of combining artificial intelligence methods with human expert knowledge is a better option that can draw on the advantages of both. In addition, to fully utilize the potential of AI in reducing overhead and complexity, corresponding design of reference signal and air interface resource allocation or even joint design between multi-link modules is required. Therefore, existing air interface frameworks and signaling Design makes more impact.

6. Plug and Play Link Control

       The 6G wireless access network needs to have the capability of automatic coverage expansion to better complete the three-dimensional full-scene coverage. When a new network service body joins the network, it can quickly shake hands, plug and play, and achieve coverage expansion. The plug and play link control technology includes the following aspects:

Process awareness: perceives various types of access requests and initiates appropriate handshake and control signaling processes. For different types of access points, it is necessary to identify accurately, complete access quickly, and realize flexible expansion of coverage.

       Cloud-to-edge control and coordination: The cloud provides flexible and precise control of edge access points, including access control, automatic allocation of bandwidth resources, and inter-link coordination. Cloud processing can introduce AI capabilities to support the above functions.

        Self-generation and self-optimization of access points: Use digital twin/AI and other technologies to fully automate and manage and monitor various access points throughout the life cycle. When the access point newly joins the network, it can automatically complete the configuration and realize self-generation; when the access point is running, it adjusts and automatically optimizes the parameters according to the real-time scene, and improves the service as needed to better meet the needs of users.

▲Plug and play link control

       High-speed and efficient transmission channels and large-bandwidth and high-real-time transmission bandwidth are required between the cloud and the edge to ensure real-time information exchange between plug-and-play interfaces. At the same time, it also requires powerful digital twin and AI algorithm support to complete remote access. point automatic control.

7. QoS control of adaptive air interface

       The 6G era will be a highly data-driven and intelligent era. New services such as holographic images, XR services, and virtual space perception and interaction have put forward more extreme requirements for the service quality assurance of 6G networks.

       The QoS control of the adaptive air interface is based on the end-to-end QoS constraints, according to the real-time air interface transmission characteristics, relatively limited air interface resources, and time constraints of transmission-feedback, etc., to realize the QoS guarantee of the air interface transmission data, which is an on-demand air interface service and Key technologies for efficient network capabilities.

       The QoS control of the adaptive air interface includes the following aspects:

       1. Flexible QoS detection mechanism: Combined with AI/big data technology, it realizes QoS detection and modeling of borne services, as well as adaptive adjustment.

       2. Deep integration of service QoS and air interface capabilities: Explore a new QoS mechanism combining service QoS and air interface service capabilities. Based on the precise requirements of the service, the radio access network matches the service requirements with the real-time air interface status through scheduling and radio resource management.

       3. The end-to-end QoS mechanism of the AS layer: the terminal combines the QoS information provided by the access network to perform more refined QoS management, so as to achieve accurate and efficient transmission of uplink and downlink data on the air interface.

Facing the future, the service requirements of 6G networks are constantly evolving. The QoS mechanism involves the core network, transmission network, and access network. Combined with the core network, the unified coordination of the QoS mechanism between the transport layer and the access network is a follow-up issue that needs to be considered.

03Network enabling technology

1. Lightweight signaling scheme

        From the development history of 2G, 3G, 4G and 5G, with the continuous expansion of the network scale and the increasing complexity, the network architecture is complex and redundant. According to the existing network development trend, the complexity of the 6G network supporting the Internet of Everything will increase exponentially. A lightweight signaling solution is an inevitable choice for 6G design.

       The 6G wireless access network needs to be designed according to a unified signaling scheme, and integrate multiple air interface access technologies under unified signaling control to achieve unified control of the air interface and reduce the complexity of terminal access to the network. In terms of protocol stack function design, differentiated protocol function design can be considered, protocol function distribution and interface design can be optimized, and the protocol function can be further enhanced by combining AI technology.

       In terms of network functions, the 6G network can be divided into a wide coverage signaling layer and an on-demand data layer. Through the separation mechanism of the signaling plane and the user plane, a unified signaling overlay layer is used to ensure reliable mobility management and fast service access; through the dynamic on-demand data layer loading, the service requirements of network users are met. Flexible cooperation between the two can reduce the number of base stations deployed and improve the user's service perception experience.

▲Lightweight signaling control

       Lightweight signaling solutions require high reliability, low latency, and low-cost transmission network support. The transmission network requires flexible topology and sufficient bandwidth. The wireless control center-transmission network-network access point needs to be integrated into the design. In addition, the separation of signaling and services requires coordinating the available 6G frequency bands to give full play to the advantages of wide coverage and flexible service loading.

2. End-to-end service design

       With the deep integration of DOICT technologies and the emergence of a large number of new services, operators need the network to have the ability to respond quickly to new demands in order to quickly provide network services. The cloud-native service-based technology is an important technology to enable the above capabilities, driving the evolution of protocol functions to a service-based architecture. The protocol function based on the service-oriented architecture has the ability to run the protocol function according to the business requirements. The technical characteristics are reflected in the following aspects:

       1. Protocol functions driven by cloud-native service technology: On the premise of complying with the logical constraints of each protocol layer, the protocol functional entities are reconstructed into flexibly combined modules. business service capabilities.

       2. Interface driven by cloud-native service-oriented technology: The internal and external interfaces of the access network are reconstructed based on the cloud-native service-oriented interface form and interface protocol, which has supported the flexible combination of protocol function modules and the opening of network capabilities;

       3. Capability opening driven by cloud-native service-oriented technology: Provide third parties with a convenient, fast and unified access network information exchange mechanism and policy adjustment mechanism to achieve a win-win situation.

▲Agreement based on service-oriented architecture

       The refactored functions of the protocol function include two categories: basic functions and incremental functions:

       1. Basic functions include cell-level functions, such as connection management, user plane management, UE energy saving management and other functions and corresponding network services.

       2. Incremental functions include access network service registration, data collection and storage, capability opening, AI analysis and decision-making, and corresponding network services.

       The high real-time performance and high flexibility of the access network function put forward high requirements on the storage, computing power and real-time performance of information interaction of the platform. Whether the DOICT deep integration technology can support this requirement needs further research and verification. At the same time, the functions of the access network are tightly coupled, and how to achieve "high cohesion and low coupling" with reasonable functions is a complex system engineering. Moreover, compared with traditional solutions, the current service-oriented technology brings an increase in the cost of a single device. How to achieve a balance between costs and benefits is a systematic problem.

      The protocol based on the service-oriented architecture runs on the cloud platform, and uses cloud native to realize the development, deployment and management based on micro-services. Cloud-native platforms need to adapt to network characteristics to achieve efficient, open, and multi-cloud deployment.

                                                                                           ▲The evolution trend of cloud technology

      In the past 20 years, computing technology has experienced rapid development from bare metal to virtual machines to containers, and cloud native has become the most suitable technical practice for cloud architecture. Cloud native is a ideological concept for cloud application design. It is the best practice path to give full play to cloud performance. It can help operators build a flexible, reliable, loosely coupled, easy-to-manage and observable network system, improve delivery efficiency, and reduce O&M complexity. Spend. Representative technologies include immutable infrastructure, service mesh, declarative API, and serverless. The cloud native technology architecture has the following typical characteristics:

     The ultimate elasticity can achieve response in seconds or even milliseconds;

      Highly automated scheduling mechanism can achieve strong self-healing ability;

      High adaptability enables large-scale, large-scale, replicable deployment capabilities across regions, platforms, and even service providers.

       Cloud native greatly reduces the threshold for cloud computing, enables cross-domain collaboration between R&D and O&M, improves the speed of open iteration, and empowers business innovation. Currently cloud native hotspot technologies are showing a blowout explosion, including multi-cloud container orchestration, cloud native server, cloud native storage, cloud native network, cloud native database, cloud native message queue, service mesh, serverless container, function as a service (FaaS) , Backend as a Service (BaaS), etc.

        Telecom services have higher requirements for performance, low latency, reliability, security, and equipment costs. These require cloud-native technologies to evolve based on the characteristics of telecom services to meet the high standards of telecom services.

3. Smart perception function

        6G-oriented ultra-low latency, high-bandwidth cloud-based interactive services are increasing. The "layered" and "chimney-like" designs of the existing application layer, service transport layer, and mobile network layer result in prolonged data packet transmission, resulting in degraded user experience.

        In order to achieve real-time and accurate matching of service transmission capabilities and network capabilities, it is necessary to introduce high-precision real-time measurement and feedback of each protocol layer of the end-to-end network to enable collaborative optimization, and introduce intelligent processing functions on the network side, including intelligent estimation and prediction. , on the one hand, it preprocesses the measurement and interaction data to achieve dimensionality reduction and compression, and on the other hand, it subscribes and notifies according to the requirements of the application layer & service transport layer to reduce the overhead of network transmission.

                                                                                         ▲ Cross-layer joint architecture design

         At the same time, it can deeply intelligently perceive the transmission requirements of the application layer, realize real-time perception and prediction of transmission requirements at the packet level under the premise of fully guaranteeing user privacy, and provide fine-grained services for congestion control at the service transmission layer and resource scheduling at the mobile network layer. Granularity guidance.

       The smart-aware network service system requires multi-protocol layers, multi-network elements, and multi-technology cooperation, and faces many challenges such as the difficulty of technical solution verification and the introduction of potential non-standard functions. At the same time, since the joint design of each protocol layer and the standardization of interaction involve multiple standards organizations and working groups, the promotion of each new technology in the standardization faces great challenges.

4. Network autonomy system based on digital twin

        Digital twin technology refers to establishing a virtual entity from the physical world entity in the digital world through digital means, thereby realizing dynamic observation, analysis, simulation, control and optimization of the physical world entity. Digital twin network technologies include functional modeling, network element modeling, network modeling, network simulation, parameter and performance modeling, automated testing, data acquisition, big data processing, data analysis, artificial intelligence machine learning, fault prediction, topology and routing Search for the best. In this way, the difficult problems that are difficult to solve at each stage of the network are converted to the digital world to solve, and the autonomous capability of the network can be realized through monitoring, prediction, optimization, and simulation.

▲Digital twin realizes network autonomy

       Based on digital twin technology and artificial intelligence technology, the 6G network will be an autonomous network with self-optimization, self-evolution and self-growth capabilities. The self-optimizing network predicts the trend of the future network state in advance, intervenes in advance for possible performance degradation, and continuously optimizes and simulates the optimal state of the physical network in the digital domain, and issues the corresponding operation and maintenance operations in advance. The physical network is corrected automatically.

        The self-evolving network analyzes and makes decisions on the evolution path of network functions based on artificial intelligence, including the optimization and enhancement of existing network functions and the design, realization, verification and implementation of new functions. The self-growing network identifies and predicts different business needs, automatically arranges and deploys network functions in each domain, and generates end-to-end service flows that meet business needs; automatically expands capacity for sites with insufficient capacity, and automatically expands capacity for areas without network coverage. Planning, hardware self-starting, software self-loading.

       As a new concept applied to the network field, digital twin technology needs to form more consensus in the industry. From the process of industry and other industries, it will take a long time. At the same time, digital twin technology relies on a large amount of data collection, which will increase the cost of equipment, and the way of data collection also requires breakthrough innovation.

5. Deterministic data transmission

       The concept of determinism was originally proposed and standardized in the IEEE. The IEEE 802.1 working group created the Audio Video Bridging (AVB) task group in 2007 with the goal of replacing HDMI, speakers and coaxial cables in the home with Ethernet. With the successful application of the IEEE 802.1AVB standard in studios, sports and entertainment venues, this technology is beginning to attract the attention of industry and the automotive world.

In 2012, the IEEE 802.1AVB Task Group was renamed the Time Sensitive Networking (TSN) Task Group. The TSN standard extends the AVB technology and has mechanisms to ensure real-time performance such as time synchronization and delay guarantee, and supports related protocols such as traffic scheduling and shaping, reliability, and configuration management. In 2015, the IETF established the Deterministic Networking (DetNet) working group to work on extending Ethernet-based deterministic techniques to wide-area IP networks, providing worst-case bounds on latency, packet loss, and jitter to provide deterministic data transmission.

        It can be seen from the above that the deterministic transmission of fixed networks has been proposed for 10 years, but the research of deterministic transmission for mobile networks has just started, mainly because 1. The air interface is easily affected by the environment, and the transmission quality is difficult to predict 2. Lack of end-to-end deterministic guarantee mechanism.

        In the 6G era, deterministic data transmission will become the representative capability of the 6G network, achieving features such as bounded delay, low jitter, high reliability, and high-precision time synchronization. Difficulties to be overcome include the following:

       1. How to realize flexible resource reservation and real-time scheduling on the wireless air interface. The unpredictability of the air interface is the main bottleneck for realizing end-to-end deterministic transmission. This requires that in the 6G era, the resources of the air interface are sufficient and unrestricted, and data packets can be flexibly scheduled in real time in the access network to ensure that the packets can be processed and sent out within the specified time.

       2. How to implement a wide-area deterministic transmission mechanism. The difficulty of applying IEEE TSN technology to a wide area is mainly due to the inability of the CNC in the TSN system to perform large-scale path operations and accurate real-time scheduling, and the time synchronization accuracy decreases as the path lengthens.

       3. How to realize cross-layer and cross-domain deterministic mechanism integration. In the 5G era, the mobile network is still a layer-over-IP network, which poses a severe challenge to the deterministic transmission scheduling of cross-domain coordination. In the 6G era, from the very beginning of network design, it is hoped to realize heterogeneous access, fixed-mobile convergence, and collaborative management. Mobile networks need to absorb the existing fixed network Layer 2 and Layer 3 deterministic transmission protocols to achieve deployment convergence and protocol support. , coordinated scheduling, so as to achieve end-to-end cross-layer, cross-domain deterministic data transmission.

6. Programmable network

       The 6G network needs to support network programmability and realize the five-network synergy of access, edge, core, wide area, and data, so that the telecom network has the ability to customize all scenarios across multiple services, multiple domains, and the entire life cycle. Network programmability is reflected in many levels, from bottom to top, chip programmability (such as P4, POF), FIB programmability (such as OpenFlow), RIB programmability (such as BGP, PCEP), device OS programmable, Device configuration programmable (eg CLI, NETCONF/YANG, OVSDB), controller programmable and service programmable (eg GBP, NEMO).

       The future network needs to meet the programmability from the four dimensions of network element, protocol, service, and management:

       1. Programmable network elements of equipment: With the diversification and individualization of data service types, users' demands for new network functions emerge in an endless stream. The network functions supported by the protocol stack of equipment network elements are limited, and the network card chips used are also limited. It is impossible to predict all possible network capabilities in the next few years. As the basic component of the network, the network element needs its hardware architecture to allow users to redefine functions, and to complete the processing of different types of protocols, encapsulation and decapsulation as needed.

        At the same time, the upper-layer software architecture consists of modules or APIs with clear functions, allowing users to reorganize these modules or call interfaces to achieve customized purposes, such as classification, shaping, and QoS. The equipment network elements support programmability, making it possible to efficiently support user customization and continuous evolution of new protocols.

        2. Programmable network protocols: The division of functions between telecommunication networks and data networks is becoming more and more blurred, and network protocols and architectures are also infiltrating each other. With the continuous evolution of application scenarios, new requirements for network protocol stack functions emerge one after another, and the evolution and innovation of network protocols (such as NewIP, SRv6, QUIC, etc.) The intra- and inter-network protocols can support synchronous handover, and even the end-to-end network protocol of the slice can be selected on demand according to the user's service type and quality requirements. And then realize the smooth switch from 5G+ network to 6G network.

       3. Programmable service paths: The end-to-end network carries more and more abundant services. We need to see that the network or network element completes the upgrade of new services in sequence. It supports on-demand configuration of different user data to use different service processing paths, which can not only adopt the old-fashioned transformation scheme, but also realize the gradual diversion to the innovative network architecture, smooth switching, and unlimited expansion to meet user needs on the basis of limited costs. Further, the forwarding path from the terminal, access network, core network, wide area network, to the entire network of the data center can be measured and adjusted, so that the end-to-end business, network, and side collaboration can be realized in a true sense, and the end-to-end network can be realized. Assure.

       4. Programmable management methods: With the increasing complexity of telecommunication networks, high intra-network O&M costs, and delays in inter-network O&M barriers, resulting in insufficient business monetization capabilities and slowing down the launch of new services. Management method course programming means that in terms of monitoring and management methods, the network elements in the network should support a variety of or customized management methods to promote the three improvements of resource efficiency, energy efficiency, and operation and maintenance efficiency, and achieve a closed-loop oriented to user experience. Autonomous network system.

       Previously, we reported the application of 5G in 21 vertical industries. With the continuous popularization of 5G, the future-oriented needs of communication will become more clear. The fields of related new businesses, new applications, new services, and new materials are developing rapidly, and new technologies and communication technologies such as cloud computing, big data, blockchain, and artificial intelligence are constantly being integrated. These urgently need to be combined with the latest changes and Development trends continue to drive 6G design and research. Although the current 6G vision may seem unrealistic, the technology often develops faster than people expect.

       This article is reproduced from "China Institute of Electronics" to support the protection of intellectual property rights. Please indicate the original source and author of the reprint. If there is any infringement, please contact us to delete