CN111292523A - Network intelligent system - Google Patents

Abstract

The invention relates to a network intelligent system which can be used as a future network paradigm. According to an example embodiment, a network intelligence system may include: the network data sensing device is arranged in one or more basic networks and used for sensing network data in the one or more basic networks; and the network intelligent processing device is used for carrying out artificial intelligent processing on the network data obtained from the network data perception device so as to obtain the information of the objects related to the one or more basic networks.

Description

Network intelligent system

Technical Field

The present invention relates generally to network architectures and, more particularly, to a network intelligence system as a future network paradigm.

Background

To address significant technical challenges of Network scalability, security, user experience assurance, real-time, mobility, manageability, and the like, developing new killer-level applications, american japan korea, and the like have successively launched a set of Future Network research plans, including the united states Internet2 (launched in 1996), the NewArch project (launched in 2000), the Clean Slate plan (launched in 2003), FIND (launched in 2006), GENI (launched in 2006), Network Science and Engineering (launched in 2009) and FIA (launched in 2010); FIRE in Europe (FutureInternet Research and evaluation, start 2007); the NWGN (New generation network, 2005 start) research Project in japan, and the AKARI Project (aka instrumentation Design Project, future light, 2006) deployed by the National Institute of information and Communication Technology (NICT) thereof; the FIRST (Future internet research for susteable Testbed, 2009) project in korea; ITU-T SG13GII (Global Information Infrastructure), FG-NGN (Focus Group on Next Generation Network, started in 2009), IMT2020Fixed/Mobile conversion, OTN (optical transport Network) beyond 100Gbit/S, FG-Net2030, etc., and a series of symbolic researches such as Internet of Things (IoT), cloud computing, big data, 4G/5G, Software Defined Network (SDN)/Network Function virtualization (Network Function virtualization, NFV) and open source, which have both revolutionary and evolutionary technical routes, play a very important role in the future Network research and have a significance in the future development of Network technology. Some of these techniques can solve the problem of insufficient addresses, but cannot solve the problem of quality of service; some can solve the bandwidth problem, but cannot solve the security problem; some heavy theories can solve the problems of local and specific services; some overweight experiments are difficult to solve the supervision problem; some are biased to be wired and fixed networks, which is difficult to solve the problem of wireless networks. The network is likely to be overwhelming if all issues and application requirements are taken into account.

From the formal investment of ARPANET in 1969 to the large-scale commercial use of global internet in the near 50 years at present, particularly in the middle and later nineties, all operator networks, all various private networks, all experimental networks, all fixed terminals, all mobile phones (2G/3G/4G/5G) and other mobile terminals are invested in over 100 trillion yuan, wherein only taking the mobile phone as an example, 74 trillion users are invested in the whole world, and each mobile phone is on average 4000 yuan, namely, 30 trillion yuan of assets. Then, backward compatibility and forward compatibility should be considered for all new technologies, new products and new solutions of future networks to protect these original 100 trillion investments. The internet is a global infrastructure of a whole course and a whole network, is a symbol of a group of 74 hundred million people all over the world, covers two hundred of countries and regions all over the world, is deeply integrated with hundreds of different industries such as education, medical treatment, industrial manufacturing and the like, has a submarine network and an aerospace network, has history and cultural deposition of nearly 50 years before and after, and is hidden by invisible hands. Therefore, the technology for using the pipe is a technology for using the pipe which can fully protect the investment of 100 trillion, and simultaneously can incorporate various new technologies which are continuously innovated in various countries, various industries and various organizations in the next decades into the future network new technology.

Disclosure of Invention

To this end, the present application proposes the concept of a "Network-Agent" (also called Network AI, NAI), which may form a scalable Network architecture, for example, it may be configured as a super Agent, an enterprise Agent, a school Agent, a hospital Agent, a traffic Agent, a provincial Agent, a military Agent, a fleet Agent, and the like. The intelligent agent is completely different from various future network research plans started in sequence such as American Europe, Japan and Korea and the like in the big idea, is not limited to the innovation of the scope of pure network, does not retrain the new method which is only provided from the concrete aspects of network system architecture, Baseline technology, service quality experience, network safety and the like to solve various problems of the network, but jumps out of the scope of pure network innovation, starts from the requirement aspect that the number of network connections is increased greatly from billions to billions and trillion of human-network-physical-object interconnection and fusion, starts from the requirement aspect that the internet is pushed from the consumption field to the industrial manufacturing field comprehensively, starts from the requirement aspect that the total human knowledge is doubled by a few decades to half a day (from the Boeing company Michael Richey), starts from the idea that the intelligent agent is compatible with the traditional network technology, protects 100 trillion investment and can be continuously brought into various new technologies, the competition centers of recent large countries and companies in the information field are not only in the range of pure networks, but also in the range of pure networks, and most of the competition centers are shifted to the information network big data and artificial intelligence fields with higher levels and wider ranges. The application provides the end-to-end and whole-course whole-network-angle human-like behavior consciousness and self-regulation capability of 'network (cloud network) + perception + big data + algorithm processing + artificial intelligence', and the problems existing in the network are mutually cooperated and jointly solved.

According to an embodiment, there is provided a network intelligence system comprising: the network data sensing device is arranged in one or more basic networks and used for sensing network data in the one or more basic networks; and the network intelligent processing device is used for carrying out artificial intelligent processing on the network data obtained from the network data perception device so as to obtain the information of the objects related to the one or more basic networks.

In some examples, the network data perceiving device includes a perceiving device for perceiving human society, a perceiving device for perceiving a natural world, and a perceiving device for perceiving a cyber space.

In some examples, the means for perceiving the cyberspace includes at least one deep packet inspection device disposed in the one or more base networks, the deep packet inspection device configured to intercept network data packets from the one or more base networks.

In some examples, the deep packet inspection device includes an inspection function, the inspection function including at least one processor and at least one memory, the memory having instructions stored thereon, the processor executing the instructions to perform the steps of: intercepting network data packets from the one or more underlying networks; parsing the intercepted network data packet to determine a service type and a related object of the network data packet; and repackaging the network data packet into a preset format based on the analysis result and sending the network data packet to the network intelligent processing device.

In some examples, the deep packet inspection device further includes an inspection management entity configured to manage operation of the inspection function entity according to instructions of the network intelligent processing device.

In some examples, the perception device for perceiving human society and the perception device for perceiving a natural world each include: an automatic sensing unit for automatically sensing the related data; and the perception control unit is used for controlling the operation of the automatic perception unit according to the instruction of the network intelligent processing device.

In some examples, the artificial intelligence process includes determining behavior and/or status of related objects based on the network data.

In some examples, the artificial intelligence process further includes determining attributes of the related objects based on their behavior and/or state.

In some examples, the artificial intelligence process further includes controlling behavior of the related objects based on attributes of the related objects.

In some examples, the artificial intelligence process further includes predicting future behavior and/or state of the relevant object through machine learning based on the network data.

In some examples, the network intelligence processing apparatus is configured to run a neural network model that is trained using the obtained network data to predict future behavior and/or state of the relevant object.

In some examples, the network smart processing device is further configured to determine an association between behaviors and/or states of a plurality of related objects through a plurality of network data related to the plurality of objects.

In some examples, the network intelligent processing device is configured to perform one or more of the following steps: determining travel information for the related objects based on the camera data, vehicle ticketing data, accommodation data; determining traffic load information for the relevant road based on the vehicle position and speed data, and adjusting the traffic control signal and/or planning the navigation path based on the traffic load information; determining bandwidth demand information for the relevant objects based on the traffic data, and configuring bandwidth and/or planning routes based on the bandwidth demand information; determining preferences for related objects based on the web browsing data and determining push content for the related objects based on the preferences for the related objects; and determining air haze data based on the haze sensor, and determining a haze pollution source based on the air haze data.

In some examples, the one or more underlying networks include a telecommunications network, the internet, and/or the internet of things.

In some examples, the one or more infrastructure networks include one or more of an enterprise network, a community network, a traffic control network, an energy control network, an environmental monitoring network, a government affairs network, a logistics network, a banking network, a school network, a hospital network, a port network, a building network, a metropolitan area network, and a backbone network.

In some examples, each base network includes a client, a server, and an intermediate connection for connecting the client and the server.

In some examples, the client comprises one or more of a personal computer, a cell phone, a tablet computer, a personal digital assistant, and a sensor, the server comprises one or more of a cloud server, a base station, a server computer, and a mainframe, and the intermediate connection comprises one or more of a relay, a bridge, a router, a switch, and a gateway.

In some examples, the one or more underlying networks include a Software Defined Network (SDN).

In some examples, the software defined network includes: an infrastructure layer comprising a network infrastructure; a control layer comprising a controller for controlling the network infrastructure, wherein the controller is defined by software; and the cooperative arrangement layer is used for performing cooperative arrangement on the network data and providing the network data to the application layer through the open port.

In some examples, the network intelligence processing apparatus controls operation of an infrastructure layer of the one or more infrastructure networks by configuring a control layer of the one or more infrastructure networks.

In some examples, the network intelligent processing device includes one or more of an enterprise intelligent processing device, a community intelligent processing device, a traffic intelligent processing device, an energy intelligent processing device, an environment intelligent processing device, a government intelligent processing device, a logistics intelligent processing device, a bank intelligent processing device, a school intelligent processing device, a hospital intelligent processing device, a port intelligent processing device, a building intelligent processing device, a metropolitan area network intelligent processing device, and a backbone network intelligent processing device, and a super intelligent processing device.

The above and other features and advantages of the present invention will become apparent from the following description of exemplary embodiments.

Drawings

FIG. 1 illustrates an architectural block diagram of a network intelligence system according to an exemplary embodiment of the present invention.

FIG. 2 shows a block diagram of the architecture of a network intelligence system according to an exemplary embodiment of the present invention.

FIG. 3 illustrates a conceptual diagram of global agents according to an exemplary embodiment of the invention.

Fig. 4 shows a block diagram of a Deep Packet Inspection (DPI) device according to an exemplary embodiment of the present invention.

Figure 5 illustrates an architectural block diagram of a Software Defined Network (SDN) according to an exemplary embodiment of the invention.

Detailed Description

The demand for network applications has overwhelmed the traditional network and network reconfiguration is imperative. From the number of users, the internet has covered more than 220 countries and regions by the end of 2017, the number of global netizens exceeds 35.5 hundred million, 76 hundred million mobile phone users, 41.9 hundred million mobile broadband users, 9.57 hundred million fixed phone users, and 9.7 hundred million fixed broadband users (from ITU data). Wireless mobile networks have evolved over millions of users, over 20 years (1980-1999), over billions of users, over 20 years (2000-2019), over billions of users including IoT, and are expected to be used for 20 years (2020-2039). Similarly, with 5000 million users, newspapers have been used for over 200 years, radios for 38 years, televisions for 15 years, and the internet for only 5 years. Network communication is one of the fields with the largest number of users, the largest industry drive and the widest influence, and the global economic, cultural, security and competitive formats are being changed profoundly. For example, currently, only the kinds of network applications APP are over 400 ten thousand, the bandwidth difference of these services is 1 Gbit/10 kbit/s-10 ten thousand times, and the delay difference is 100ms/10 ns-1000 ten thousand times. The number of users and the types of services are increased dramatically, the difference is huge, the traffic pressure is not enough, the expandability, the safety, the user experience guarantee, the real-time performance, the mobility and the manageability of the network are challenged unprecedentedly, but the network is adapted passively, and the network reconstruction and the introduction of intelligence are really necessary and urgent.

In 2017, the total amount of digital economy taking the internet as an important carrier in China has reached 27.2 trillion yuan and accounts for over 32.9 percent of GDP (white paper book for development and employment (2018)) in China. The electronic commerce transaction amount of China accounts for 42 percent of the whole world, and the mobile payment transaction amount is 11 times of that of the United states. The digital economic scale of China is expected to exceed 32 trillion yuan in 2020. Almost all industries are networked, but networks are passively accommodating. By the beginning of the 17 th century, the income of all middle-aged and middle-aged people in western economy is doubled, and the income is used for 800 years; subsequently, the per-capita income increased by 13 times, and only after 150 years, it was seen that the economy itself developed more and more rapidly. The american economist Brian Auther recently proposed the concept of the second economy, namely, the economy running on and in the processor, the sensor, the network device, the mobile phone, and the other economic activities, which forms the second economy beyond the well-known physical economy (the first economy), and is equivalent to the "nerve layer" above the first economy, so that the national economy is informationized and intelligentized. This is the largest change over a hundred years of electrification. Brian auth also believes that by 2035 years ago, the second economy will scale toward the first economy. The convergence of the network and various industries is further aggravated, but the network pressure is huge, the ubiquitous availability of network access, the convenience of use and the reliability and stability of a basic network are all greatly challenged, the network is passively adapted at present, and the network reconstruction and the introduction of intelligence are really necessary and urgent.

From 5G and 3 large application scenes of a license plate to be issued, a typical value of user experience rate is 0.5-1.0 Gbit/s, the connection number density is millions of users per square kilometer, the peak rate is 10Gbit/s, the flow density is 10Tbit/s per square kilometer, the mobility is more than 500km/h, and the network delay is less than 1 ms. The communication from person to person can be expanded to person to object, object to object, Internet of vehicles and the like, and the number of connections can be expanded from the current billion level to the billion level. The spectrum of 5G will be much higher than 4G and the density of base stations will be more than 3 times that of 4G. In 2017, the number of 4G base stations reaches 328 ten thousand, so that the number of 5G base stations only enhancing a mobile Broadband (eMBB) type can finally reach more than 1000 ten thousand by 3 times of calculation. The dramatic increase of the number of new high-bandwidth services can lead to the exponential increase of the flow of core networks such as metropolitan area networks, provincial trunk networks, national trunk networks and the like, the pressure of user experience is increased by several times, the reliability and stability of the core networks are greatly challenged, the introduction of edge calculation brings some complexity to the networks, the deployment of IoT increases the difficulties such as wide coverage, low power consumption and the like, the networks can continuously adapt passively, and the network reconstruction and the introduction of intelligence are really necessary and urgent.

The global network extends from the consumer domain to the manufacturing domain, europe promulgates a digital industry action plan, germany industry 4.0 promulgates a digital strategy 2025, france promulgates a new industry france plan, the uk promulgates future manufacturing forecasts, the united states promulgates an advanced manufacturing strategy industry internet, and japan promulgates a manufacturing innovation 3.0 strategy and a new robot strategy. All countries take manufacturing industry as the core and extend the manufacturing industry to physical economy. China has more than 220 industrial product yields worldwide first. After the implementation of Chinese manufacturing 2025, the production mode is changed greatly, from large-scale production to personalized customization, from production type manufacturing to service type manufacturing, from element driving to innovation driving, and intelligent production, network cooperation, personalized customization and service extension are realized. However, the network is to take the industrial internet to bear a heavy burden, and all raw materials, finished products, intermediate parts and all manufacturing equipment of all factories around the world are all connected together, technical flow, fund flow, logistics (spare parts, intermediate parts and finished products), information flow and people flow are mixed, and the connection number and the type exceed the current number of people-to-people connection. The extensibility, user experience assurance, real-time performance, mobility and manageability of the network are challenged unprecedentedly, and the challenges of device security, network security, control security, application security, data security and the like are also faced, the network is being adapted passively, and the network reconfiguration and the introduction of intelligence are really necessary and urgent.

The reports of the McKensin consultant company indicate that the industrial revolution on which we are coming is 10 times as fast as the social revolution caused by the first industrial revolution (steam engine initiation), 300 times as fast as the scale and 3000 times as fast as the influence, but the network is passively adapted. In the past 10000 years, the evolution speed of human genes is almost 100 times faster than that in the previous 6000000 years, and the human body is improved while tools are improved. One expert of Boeing company mentions the time for doubling the total amount of human knowledge in a lecture of 2017 in 9 months, the time for doubling in 1440-1900 years, the time for doubling in 1900-1950, the time for doubling in 1950-1970, the time for doubling in 2017 in 13 months, the time for doubling the total amount of knowledge is expected to be shortened to 12 hours in 2020, the time for doubling the total amount of knowledge is shorter and shorter, the total amount of big data is larger and larger, the pressure of data processing capacity is increased continuously, various types of data are generated by a machine, read by the machine, analyzed by the machine and counted by the machine at the moment, the data processing capacity of the network machine is passively adapted, and the network reconstruction and the introduction of intelligence are really necessary and urgent.

From the global development situation, the network technology is in the important opportunity period of reconstruction innovation and intelligent introduction, the network infrastructure is accelerating to develop towards the direction of large bandwidth, wide coverage, cloud network integration and intellectualization, the network threat and the traditional means are linked together to become an important source influencing the national security, the network space becomes the strategic space gaining the long-term competitive advantage of the country, and the foundation of the urban intelligent agent is supported by equipment intelligence, network intelligence and business intelligence triangles. By 2040 years, the internet was about 30000 times larger in size than the current internet if continued to grow without major consequences. By then the "network" will not be centered on the network itself, but on the traffic and its intelligence (as with Tencent, Ali, etc. driving network traffic far more than the network equipment manufacturers).

From the value of the network, the mertecavir law expresses the network value as: value of network and number of linked nodes within systemIs proportional to the square of, V ═ k × N²In the formula: k is a value coefficient; n is the number of networking nodes (the number of world man-machine networking devices has grown from 0.08 in 2003 to 1.84 in 2010, 3.47 in 2015, and possibly 6.58 in 2020). The network is a national core resource, the architecture is soul, and the innovation is a strategic measure. And (4) conclusion: the number of users, industrial drive and influence at the current stage of the network are unprecedented; the problems, difficulties and challenges that exist have not been previously addressed; the future development-oriented tasks are heavy, high in pressure and hope, the available technical means are limited, and the time window is also very limited.

In view of the technical problems and challenges of existing networks, the devices in the network have a large number of complex closed systems and a large number of private/internal interfaces. The network lacks flexibility, the expansibility of the device function is weak, a large number of service islands are formed, the services are difficult to fuse, new services are difficult to develop, the operation and maintenance are complex, and the operation cost is high. The exterior faces OTT challenge, the interior faces business innovation difficulty, growth weakness, business volume and business income scissor difference continuously expanding.

In summary, the network reconfiguration is necessary and is only the first step, and the flexibility, the intelligentization strength and the introduction of the network agent are further increased subsequently, so that the device intelligence, the network intelligence and the service intelligence are realized, and the network agent is supported and formed together.

FIG. 1 illustrates an architectural block diagram of a network intelligence system according to an exemplary embodiment of the present invention. As shown in fig. 1, the network agent system 10 may include a sensing layer 11, a network 12, and a network agent (NAI, or agent for short) 13, where the network agent 13 may include a cloud server agent 13a and an intelligent command hub 13 b. Taking a city as an example, after the facilities are deployed, the whole city is like a human body, and is coordinated, self-perception, self-statistics, self-reasoning, self-learning and self-prediction through the brain finger-waving of hands, feet, hearing, speaking, smelling, taste, skin and nervous systems to form an organic whole.

The sensing layer 11 may include various network data sensing devices for sensing various network data and transmitting the sensed data by means of a network. Thus, the awareness layer 11, or network data awareness, may itself be deployed in the network. For example, the sensing layer 11 can sense data related to human society, the natural world and network space, and in short, the sensing layer 11 can sense data generated by human-network-object interconnection and fusion. Due to the development of technologies such as the internet of things, the industrial internet, the internet of vehicles, the ocean network, the air-to-air network and the like, the data amount which can be perceived by the perception layer 11 is larger and larger, and almost most fields can be covered. If necessary, the sensing layer 11 may also perform necessary processing on the sensing data, such as parsing to identify the service type, repackaging into a predetermined format, etc., and then transmitting to the upper layer device.

The human society may include data related to human activities such as traffic data, data generated by human use of personal computers, cell phones, etc. (e.g., web browsing data, WeChat click documents, web shopping), etc. For example, traffic data may be generated by a traffic camera, a car networking, and the like. The natural world may include, for example, raw materials purchased by a factory, production equipment, middleware produced, products produced, warehouses, outdoor temperature, climate, air quality, rainfall, remote videos of scenic spots, etc., which may be generated by various sensors or RFID. The network data includes various data transmitted in the network (such as capital stream data, physical distribution data, management data, voice, video, picture data, etc. of the plant). The sensing layer 11 can sense these data by various means (note: not infringing the privacy of the individual, operating within the allowed scope of the laws and regulations). For example, the means for perceiving cyber-spatial data of the perception layer 11 may include a Deep Packet Inspection (DPI) device deployed in the network to intercept network packets transmitted in the network, as will be described in further detail below. Alternatively, the means of the sensing layer 11 for sensing the human society and the natural world may include an automatic sensing unit 11a and a sensing control unit 11 b. The automatic sensing unit 11a may be configured to automatically, e.g., continuously or periodically, sense data generated by activities of the human society. For example, a camera may capture personal movement data, traffic data, etc., a microphone may capture voice data of a person, etc., and data generated by the action of a person tapping a keyboard, clicking a mouse, etc., may also be captured. Also, the auto-perception unit 11a may be configured to send perception data to the network agent 13 via the network 12. As will be described below, the network agent 13 may determine information about the relevant objects based on the received network data and may also control the data-aware process through instructions. For example, the network agent 13 may send an instruction to the sensing control unit 11b, and the sensing control unit 11b may control the operation of the automatic sensing unit 11a according to the instruction, for example, control the data traffic type sensed by it, the sensing frequency, and the like. It should be understood that the network data aware means may be implemented by software, hardware or a combination thereof.

Although not shown, the network 12 may include various networks, such as one or more underlying networks that are different from or the same as one another, examples of which include, but are not limited to, telecommunications networks (including 2G, 3G, 4G, 5G networks, and future developed telecommunications networks), the internet of things, local area networks, and so forth. Examples of one or more underlying networks include, but are not limited to, enterprise networks, community networks, traffic control networks, energy control networks, environmental monitoring networks, government affairs networks, logistics networks, banking networks, school networks, hospital networks, port networks, building networks, metropolitan networks, and backbone networks, among others, in terms of functional characteristics. As will be understood from the detailed description below, in one or more underlying networks, through various terminals and servers, etc., a multitude of network data may be generated, which is transmitted within the network as well as between the networks.

The network agents 13, which may also be referred to as network intelligence processing means, may perform artificial intelligence processing on the network data obtained from the perception layer 11 to obtain information about objects associated with the underlying network. Here, the object related to the infrastructure network may include the human society (including people) described above, the natural world (e.g., raw materials purchased by a factory, production equipment, middleware for production, products for production, warehouses, outdoor temperatures, climate, environment, etc.), or the network itself (e.g., various network devices), and the like.

Network agents 13 may include cloud agents 13a and an intelligent command hub or center 13 b. The cloud agent 13a may be deployed on the network 12 as a cloud, and receive network data transmitted through the network 12 from the perception layer 11. The cloud agent 13a may perform necessary processing on the network data, including data fusion, classification, statistics, storage, and the like, or may also include simple reasoning, association, and the like. For example, the cloud agent 13a may store data related to a certain object, such as a person, in a classified manner to hold information related to the person. The intelligent command hub 13b may perform further artificial intelligence and algorithm processing based on the data of the cloud server agent 13a, including self-perception, self-statistics, self-inference, self-learning, self-prediction, etc., which may be implemented by means of software and hardware related to the field of artificial intelligence, which are currently available or developed in the future. Some functional examples of the network intelligent processing device 13 will be described in further detail below.

FIG. 2 shows a block diagram of the architecture of a network intelligence system according to an exemplary embodiment of the present invention. By way of example, two base networks 10a and 10b are shown in FIG. 2, but it should be understood that a network intelligence system may include more or fewer base networks.

Each base network 10a and 10b may include clients, servers, and intermediate connections for connecting the clients and servers. In the example of fig. 2, the base network 10a may be, for example, the internet (or the internet), wherein the clients may include, for example, a tablet computer 14, a personal computer 15, and may also include, for example, a personal digital assistant, and various sensors (not shown), such as a haze sensor, a water quality sensor, a traffic camera, a security camera, an RFID sensor, a barcode scanner, a magnetic coding sensor, etc., which are connected to the network, which are not exemplified herein. The intermediate connections may include, for example, switches 16, and may also include, for example, repeaters, bridges, routers, gateways (not shown), and the like, used to connect various network devices, including clients and servers, for data transfer. The server may include, for example, the server computer 17, and may also include, for example, a mainframe, a super computer, and the like, and of course, the server may also be deployed as a cloud server cluster. The base network 10b may be, for example, a telecommunications network, wherein clients may include, for example, handsets 18, servers may include, for example, base stations 19, network management servers (not shown), etc. It should be understood that the base networks 10a and 10b are only examples and the present invention is not limited thereto.

It will also be appreciated that the network elements given above as examples may serve as one or more of a client, a server, and an intermediary connection. For example, the base station 19 is used to connect different handsets 18 together, so the base station 19 can also be considered an intermediate connection. The personal computer 15 may also be configured as a server to provide certain services, such as HTTP web services, and thus may also act as a server. For another example, a mobile phone, a personal computer, etc. may also be regarded as a sensor for sensing operations, behaviors, etc. of a user thereof, and thus, in a broad sense, various terminals in a network may be regarded as a sensor.

As an example of the network data perceiving device 11 for perceiving the network space, a Deep Packet Inspection (DPI) device 21 is shown in fig. 2. One or more Deep Packet Inspection (DPI) devices 21 may be deployed in the infrastructure networks 10a and 10b, for example, near an intermediate connection or network data transmission path (e.g., fiber, cable) to intercept network data packets and send the intercepted network data packets to the network intelligent processing device 30. The Deep Packet Inspection (DPI) device 21 will be described in detail below. Although a Deep Packet Inspection (DPI) device 21 is shown here as an example, it should be understood that the network data awareness device 11 for awareness of network space may be implemented in other ways. For example, the network data aware device 11 may be an intermediate connection in the underlying network or a part thereof for copying and sending network data packets forwarded therethrough to the network intelligent processing device 30. As another example, the network data aware device 11 may be a software defined controller that controls network elements in the underlying network to send a copy of network packets it generates, receives, or forwards to the network intelligent processing device 30.

The network intelligent processing device 30 may include one or more sub-processing devices respectively for performing artificial intelligence processing on network data of different underlying networks. For example, fig. 2 shows a sub-processing device 31 for the infrastructure network 10a and a sub-processing device 32 for the infrastructure network 10 b. In addition, the network intelligent processing device 30 may further include a processing device 33 capable of processing network data of a plurality of infrastructure networks. The processing devices can be classified into an enterprise intelligent processing device, a community intelligent processing device, a traffic intelligent processing device, an energy intelligent processing device, an environment intelligent processing device, a government affair intelligent processing device, a logistics intelligent processing device, a bank intelligent processing device, a school intelligent processing device, a hospital intelligent processing device, a port intelligent processing device, a building intelligent processing device, a metropolitan area network intelligent processing device, a backbone network intelligent processing device, a super intelligent processing device and the like according to functions thereof, wherein the backbone network intelligent processing device can process data of a national network, and the super intelligent processing device can process data of a global network. Each intelligent processing device may be used to implement the cloud server body 13a and the intelligent command hub (or brain) 13b shown in fig. 1.

Here, the network intelligence processing apparatus may also be referred to as a network agent, which determines information of objects related to the network, including behaviors and/or states of the related objects, and the like, based on artificial intelligence processing of network data. The network intelligent agent is a brand new intelligent agent with a brand new form, which is based on network and human-network-object three-element mutual connection, has a relatively complete networking and intelligent system, has certain self-perception, self-learning, self-decision, self-execution and self-adaptation capabilities, and has certain mind and human-like behavior consciousness. The system, the platform or the network which can be materialized can be used for realizing (or embodying) the entity which is continuously upgraded and evolved, has timeliness and weak central control and high connectivity manually, and the entity needs to cooperate and interact with 'network (cloud network) + perception + big data + algorithm processing' to solve various problems in the future network together. The conversion is from sensing to data, from data to information, from information to knowledge, from knowledge to intelligence, from intelligence to service, the network is still neutral, and the privacy of the user needs to be protected to ensure the information security.

The implementation of network agents can take full advantage of the hardware and software levels that have developed now, including the hardware and software associated with artificial intelligence that has developed rapidly in recent years, as well as quantum computers that are developed in the future, and the like. In particular, the rapid development of artificial intelligence has given machines the ability to perceive and learn, which has the potential to break through "singularities" in machines, i.e., beyond the capabilities of humans. With artificial intelligence, human-like perception, learning, and prediction processes have been enabled by machines. For example, with artificial intelligence processing, in addition to determining the behavior and/or state of a related object based on network data, attributes of the related object may be determined based on its behavior and/or state, and its behavior may be controlled based on the attributes of the related object. For example, it may be determined that a network user performs network attack many times based on network data, and thus it may be determined that the network user is a hacker. When the user is determined to be a hacker, the network agent may restrict the network operation of the user by configuring the network, in particular to restrict its network attack behavior, e.g. may cause its attack instructions not to be correctly routed to the target network element or address. Conventional networks can only resist network attacks by improving the security of the network itself, but it is theoretically almost impossible to build a network completely free of security holes. The network of the invention can limit the attack behavior of a user by actively determining the user as a hacker, thereby being capable of solving the problem of network attack more effectively. As another example, based on the energy consumption data and the environmental haze monitoring data obtained from the internet of things, it may be determined that a certain object (e.g., a factory or a site) is a haze pollution source, and then the energy consumption (e.g., power distribution) of the object may be limited to reduce the haze pollution. As another example, the network traffic data of a certain user may determine the network bandwidth required by the user, so that the user may be configured with the required bandwidth. The service can be applied to a large number of users, thereby achieving reasonable and efficient configuration of network bandwidth.

In some embodiments, the artificial intelligence process of the network agents may also utilize recently developed machine learning techniques to predict future behavior and/or states of related objects through machine learning of existing data. For example, the network intelligent processing device may run a neural network model established for a specific application scenario, train the neural network model using the obtained network data, and then predict the future behavior and/or state of the relevant object using the trained model. The model may also be continually modified using truth data during the prediction process. For example, future data may be predicted from historical data of traffic, weather, and the like.

In some embodiments, the network smart processing device may also determine an association between behaviors and/or states of a plurality of related objects through a plurality of network data related to the plurality of objects. For example, the main emission source may be determined by determining the relationship between the nationwide haze distribution and the density distribution of emission sources of factories, vehicles, and the like. The network agents may utilize the big data obtained from the network to stand at a higher perspective to view the network related objects, thereby making it easier to discover associations between the objects.

Some examples of specific applications of the network intelligent processing device are described below.

Example 1: and determining the travel information of the person based on the camera data, the vehicle ticketing data and the accommodation data, and realizing the tracking of the person. The network camera can shoot pedestrians in various scenes to obtain face images of the pedestrians. In some embodiments, facial features can be extracted by software or dedicated hardware such as a chip at the camera head end by using edge computing technology, and then transmitted to the network intelligent agent for identification and matching, which can greatly reduce the computing power requirement of the intelligent agent end and reduce the network bandwidth consumption. Further, traffic ticket data such as railway, airplane, ship ticket purchasable information, accommodation information, and the like can provide travel information of a person as a specific object. The network intelligent agent can track the journey of a person by extracting and integrating the information.

Example 2: traffic load information for the relevant road is determined based on the vehicle position and speed data, and traffic control signals are adjusted and/or a navigation path is planned based on the traffic load information. The construction of the 5G network promotes the development of the vehicle networking, the networked vehicles can send information such as the positions and the speeds of the networked vehicles to the traffic management intelligent agent, and the traffic management intelligent agent can determine the conforming information of the traffic network through the data, so that traffic control signals can be adjusted and/or navigation paths can be planned. For example, if the load in the east-west direction of a certain intersection is much higher than the load in the north-south direction, the green light time in the east-west direction can be prolonged, and the street light time in the north-south direction can be shortened, so that the traffic flow is accelerated, and the traffic pressure is relieved.

Example 3: bandwidth demand information for the relevant objects is determined based on the traffic data, and bandwidth is configured and/or routes are planned based on the bandwidth demand information. For example, if the network traffic of a certain terminal is large, a large network bandwidth can be configured for the certain terminal. For another example, if the load of a certain routing path is large, a new routing path can be planned, and faster access is realized. One advantage of the network agent is that the whole network can be observed globally, so that the optimal configuration of resources is realized, and the utilization efficiency of the network is improved.

Example 4: preferences of related objects are determined based on the web browsing data, and push content for the related objects is determined based on the preferences of the related objects. For example, if the browsing content of a certain user is related to the clothing accessories and the like, popular element information such as the clothing and the like can be pushed for the user; if the browsing content of a certain user is related to finance, finance news or products and the like can be pushed for the user.

Example 5: the air haze data is determined based on the haze sensor, and the haze pollution source and the like are determined based on the air haze data. This example has been described previously and will not be repeated here.

The above are only a few examples, and it should be understood that a network agent may have various functions depending on the network it serves, which may be implemented using existing levels of hardware and software.

From the perspective of global intelligent agent, it is cheap, and is formed from optical fibre, radio electromagnetic wave, copper wire, silicon atom, bit, software and algorithm, etc., and can consume the electric energy of tens of large-scale nuclear power stations on earth, so that it is an unimaginable huge networkEverywhere is absent, no hole is absent, and the whole world is covered by 5.1 multiplied by 10¹⁴The range of square meters, which relates to land, sea, air and space and hundreds of large and small industries including education and medical treatment, is the largest system established by human beings at present, and the influence and the penetrating power on daily life are ten to one hundred times larger than the industrialization.

From the class of Network Agents (NAIs), they can be very large, such as the global agents mentioned above; can be large-scale, such as national agent (can be divided into a plurality of agents according to functions, such as education, science and technology, national defense, import and export and the like), for example, the modern political information agent appears, if can be utilized, the treatment capacity, level and efficiency of the state can be many times higher than the current; may be moderate, such as city agents, provincial agents, operator agents, national traffic agents, educational agents, cross-country company agents, Tencent/Ali agents, and the like; the system can also be small-medium-sized, such as an enterprise agent (a super enterprise, compared with the current 'enterprise + informatization', the competitiveness is increased by a large level, and the efficiency can be improved by several times to dozens of times). The 2000 data shows that of the 100 largest economies worldwide, 51 are corporations and only 49 are nations. Several years later if they become intelligent, the impact of their decisions is far and widespread globally and of great importance.

People click thousands of billions of times through mobile phones and other terminals every day to tell the NAI what the human is doing, what the human may do next, and the NAI completes functions such as intelligent learning and the like. The method comprises the steps of continuously learning every day, continuously carrying out intelligent statistics and classification on big data, carrying out deep mining, continuously finding internal relevance (for example, finding a haze source through a batch of isolated and dispersed haze data, carrying out case detection and rapidly finding suspects through a batch of seemingly irrelevant public security video information, and the like), and upgrading to new intelligence. The study characteristics are defined as follows:

(1) the theory, the method and the application of 'cloud network + perception + big data + algorithm + artificial intelligence' are researched and developed on the basis of the Internet, network reconstruction and SDN/NFV and network big data thereof, so that the proper use of the artificial intelligence is realized.

(2) A new intelligent entity (composed of software and hardware, such as software upgrading of an SDN/NFV controller) capable of reacting in a manner similar to human intelligence can be developed and produced, and research in the field comprises network self-perception, self-search, machine self-learning, logic self-reasoning, certain self-thinking capability and expert system capability generated through training and the like, so that the principle of intelligence is realized, an algorithm similar to human brain intelligence is designed, and a computer can realize higher-level intelligent application.

(3) The system can complete a large amount of repetitive primary mental labor like human beings, has certain thinking capability in the aspects of data correlation, timeliness and individuation exceeding the human beings in the big data processing, and can greatly exceed the intelligence of the human beings in the later period. Some of the more complex tasks that normally require human intelligence can be accomplished through training. The understanding of this "complex work" by different industries, different locations, and different groups of people is different.

The connection quantity of NAI can be rapidly developed from billions to billions and trillions, from human-network binary interconnection to human-network-object ternary interconnection and even multiple interconnection and fusion, from ground plane interconnection to space three-dimensional interconnection and outer space and interstellar interconnection (NAI is equivalent to adding a nerve layer or control layer on the traditional physical world and human society, and is deep informationized in nature), and the total quantity of network data can be rapidly developed from TB and PB levels to EB, ZB, YB and even BB levels. The method greatly breaks through the limitations of the self ability and the mental power of human beings, greatly breaks through the limitations of space and time, and greatly breaks through the limitations of a large amount of human facilities and resources such as a worldwide library, a large encyclopedia, a worldwide university, a worldwide hospital, a worldwide movie theater, a worldwide factory, a worldwide shop, a worldwide post office and the like.

From NAI worldwide, a huge intelligent agent which is spread all over the world, land, sea and air, has no time and place, can be used at the grassroots level at any time and any place, has no holes and can not be input, has concentrated on the intelligence of 74 hundred million people, is dynamically connected with 74 hundred million people, changes ' loose ' connection of 74 hundred million people into ' tight and dynamic ' connection, changes ' billion isolated ' dead ' objects on the earth into ' live and dynamic connected ' objects, is a dynamic artifact which is the biggest history of human beings, is the most complicated and hard to imagine event which is not transferred by human intention and is artificially produced by the earth so far, and is composed of silicon materials (hardware), equipment, 0/1 codes (software), networks, algorithms, data and the like (the networks are still neutral, the user privacy must be protected, and the information security problem always troubles us people). The human brain is composed of approximately 800 million neurons, NAI can be at least billion-fold more than human neurons, and its capacity, amount of knowledge, and intelligence are accelerating to increase. The NAI has better adaptability and can perform self-adjustment on external changes in a very wide range; the method has the evolutionary capability, and can transfer the adaptability obtained by local construction from one region to another region and from one application category to another category; the NAI is established on a plurality of parallel relations, redundancy exists, and even if a large fault occurs, the NAI is equivalent to a small fault in a higher level; the expandability is realized, a more complex new structure can be constructed, and more useful information can be bred; with novelty, the combination of small individuals associated with each other in a large system grows exponentially. Individual differences are allowed in the cluster, which by evolution can lead to constant newness.

The amount of economy driven by it will approach and exceed the physical economy, which is a necessary trend. It is expected by the agency that the data usage by the 2020 world-web-object interaction and fusion will reach about 44ZB (1 ZB-10)¹²Megabyte), will cover each field of economic society development, will reconstruct links such as our traditional production, distribution, exchange, consumption, will change our employment structure, impact law and social ethics, infringe individual privacy, to government management, economic security, social stability and global governance etc. produce far-reaching influence, national security and international competition will be more complicated, will produce new uncertainty. The generated revolutionary influence can remold the productivity development mode again, reconstruct the production relation, improve the industrial efficiency and the management level, and deeply change the production life style of human beings andthe thinking mode can continuously improve the accuracy, the efficiency, the predictability and the service level of government governments. These big data will create next generation internet ecology, next generation innovation systems, next generation manufacturing modalities and next generation social governance structures.

A typical example of a network agent is a city agent serving a metropolitan network, the concept of which is broad and well understood. It contains political, economic, cultural and technical contents. The concept of IoT was formally proposed by international union in 2005 on the internet basis; IBM proposed the intelligent Earth's Fairy in 2008, and received much attention. IBM in "City of wisdom in China" white paper, defines "wisdom city" as such a city: the city operation core system can sense, analyze and integrate various key information of the city operation core system by fully utilizing information and communication technical means, thereby responding to various requirements including civil life, environmental protection, public safety, city service and industrial and commercial activities intelligently and creating better city life for human beings. Its 4 characteristics are comprehensive things, full integration, motivation innovation and cooperative operation. In contrast, china is the biggest developing country in the world, and "development, governance, society, life" is still the subject of the high-quality development of the urbanization in china in the next decades, and should be included in the definition.

The city intelligent agent is defined by combining the national conditions of China as follows: the city takes 'more scientific development, more efficient treatment, better life and more harmonious society' as the overall target of high-quality development, has a complete digital, networked and intelligent system, has certain self-perception, self-learning, self-decision, self-execution and self-adaptation capabilities, is a brand new city with the urban intelligence and humanoid behavior consciousness, and is an upgraded version of the smart city at the higher stage at the present stage. The method needs to cooperate and interact with 'network (cloud network) + perception + big data + algorithm processing', and various problems proposed in future network and city development are solved together.

Taking an urban traffic intelligent agent as an example, the urban traffic intelligent agent refers to the whole-course whole-network dispatching and commanding intelligentization of the whole urban traffic, and each intersection has certain dynamic self-sensing, self-learning, self-decision, self-regulation and control capabilities of traffic flow and routing detouring capabilities. The driver can know the traffic jam condition at any time through the terminal device in the vehicle, and can also obtain the traffic jam forecast, and the traffic accident information can be automatically sent to each vehicle. The parking lot conditions near the shortest path and the destination are automatically obtained, the parking space is reserved, and the vehicles are automatically sensed and are subjected to networking accounting through a direct bank. The robber can alarm at any time when encountering the robber. Once the vehicle is stolen, the position of the vehicle can be known at any time, and if the terminal in the vehicle has important potential safety hazard, the terminal can prompt in time. Emergencies and important weather changes are also automatically prompted. The traffic information and flow, annual inspection of vehicles and violation information are automatically counted and classified at regular time, and the switching time of traffic lights at the crossing is automatically and dynamically adjusted.

The construction and development of urban intelligent bodies can bring forth a new intelligent industry, and the intelligent industry is taken as a link to promote the self-benign operation of the whole city. The intelligent industry brought by the urban intelligent agent can be understood as follows: based on the physical world and the physical society of a real city, information, knowledge and mental resources are used as supports, and information processing technology is utilized to carry out deep intelligent analysis and refinement on various things so as to realize a novel industrial system (such as online tour, Facebook, virtual community and the like) with relatively perfect self-perception, self-learning, self-decision, self-regulation and control capability and behavior awareness. It is different from the information industry as well as the knowledge industry. The intelligent industry can not only form huge virtual Space different from network Space (Cyber Space) and enlarge intangible scale of cities, but also comprehensively develop and utilize the intelligent virtual network Space.

The difference between the city intelligent agent and the city informatization lies in that:

(1) the urban intelligent agent takes 'development is more scientific, management is more efficient, society is more harmonious, life is more beautiful' as a high-quality development target, and the whole city has more perfect behavior awareness and self-regulation capability and has space-ground multi-platform cooperative processing capability;

(2) the whole city has a relatively complete digitalization, networking and intelligentization system and has certain intelligent sensing, situation sensing and cognition capabilities;

(3) the whole city has a mature information-knowledge-intelligent self-conversion mechanism, and a certain city intelligence and human-like behavior awareness;

(4) the whole city has certain self-learning, self-growing and self-innovation abilities and the like.

The city intelligent agent integrates the informationized system, the IoT system and the community network which are dispersed and are respectively administrative in the city, and is promoted to be an organic whole with better coordination capability and self-regulation capability, which is not existed before, is sublimation and leap of the smart city in the traditional sense, and is endowed with new connotation.

And a simple comparison with the digital government. The positioning of the digital government is to make a scientific decision, fine management and high-efficiency service digital government, which means that government organizations apply modern information and communication technology, the management and service functions are fused through network technology, the optimization and recombination of government targets, organizational structures and work flows are realized on the Internet, the boundaries among time, space and departments are broken through, and high-quality, all-round, standard and transparent management and service which accord with the international standard are provided for citizens. The digital government has the same characteristic points with the intelligent government in the aspects of 'applying modern information and communication technology', 'realizing the optimization and recombination of government targets, organizational structures and workflow, breaking the division boundary' and 'providing high-quality and all-round management and service to the society', different characteristics or different characteristics of urban intelligent agents and urban informatization, and the 4 items, for example, 'permeating the functions of the intelligent agents to each corner of a city through self-perception, self-detection, self-learning, self-decision and the like' are not related to the digital government.

The understanding of the city intelligence described herein is different from the "developed national wisdom city solution and IBM description". For example, the understanding, positioning, function, structure, image and characteristic, cultural habit, political mode and treatment mode of the city essence in China are completely different from the system of western developed countries, so that the positioning, starting point and requirement of urban intelligent agents in China cannot be moved to the foreign mode. In addition, each city in China has already completed a large amount of highly effective stage work in the aspects of digital government, city informatization, digital city and the like, plays a positive promoting role in city development in China, cannot lose the important factor, and needs to be inherited and developed. Therefore, a solution should be provided based on the specific national conditions of China. The realization of the urban intelligent agent can further liberate and develop productivity, liberate citizens from simple and repeated labor, realize the change of government management from extensive to fine, the change of singleness to cooperativity and the change of passivity to initiative, and really realize decision scientification, treatment accuracy and high service efficiency.

The urban intelligent agent is a product combining virtual economy and entity economy, can promote production, life, management modes and economic social development in an urban area to have unprecedented deep changes, can reduce and save the investment of various substances and energy sources in cities to a great extent, reduce the consumption of resources and energy sources, reduce urban environmental pollution, further improve the effect of market allocation resources, further improve labor productivity, and provide a high-quality-development urban road with high technological content, good economic benefit, low resource consumption, less environmental pollution and fully exerted human resource advantages.

The NAI and the urban intelligent agent are all intelligent agents, and are all end-to-end and whole-course full-network entities with human behavior awareness and self-regulation capability, namely 'network (cloud network) + perception + big data + algorithm processing + artificial intelligence'. The differences are as follows: NAI emphasizes more on solving the network problem of end-to-end and whole-course whole network, emphasizes various group intelligent behaviors based on network and human-network-object three-element mutual interconnection, emphasizes establishing a relatively complete networking and intelligent system, emphasizes establishing an intelligent agent with certain self-perception, self-learning, self-decision, self-execution and self-adaptation capability and certain intelligence and human-like behavior awareness in city range. The city intelligent agent emphasizes more in the city range, solves the network problems of end-to-end city range and whole-course whole network, emphasizes more cities, takes 'development is more scientific, governance is more efficient, life is better, society is more harmonious' as a high-quality development overall target, builds a complete digital, networked and intelligent system, builds a city with certain self-perception, self-learning, self-decision, self-execution and self-adaptation capabilities and a brand new form of city intelligence and humanoid behavior consciousness, and is an upgraded version of the smart city at the higher stage at the present stage.

Another typical example of a network agent is a global agent serving a global network, or may also be referred to as a super agent, of which fig. 3 shows a conceptual diagram. The global agent can collect information of the global network, extract information behind the data through big data and artificial intelligence processing and the like (note: the privacy of the individual must not be invaded, the operation is allowed by the corresponding laws and regulations), observe and understand the world beyond the global view of the whole human, so that the world can be more comprehensively and deeply understood, and the global development is promoted and optimized. Global agents can cover the world-wide network area, including marine, land and space networks, and relate to various industries, including education, medical services, etc., which will be the largest system established historically by humans, with 10-100 times greater impact on human daily life than the industrialized revolution. The global agent can be used at any place and any time, benefits the world with 74 hundred million people, changes loose connections between people into tight and dynamic connections, changes hundreds of billions of isolated affairs on the earth into mutual-related affairs, and profoundly changes the political economic culture life of human beings.

Fig. 4 shows a block diagram of the Deep Packet Inspection (DPI) device 21 shown in fig. 2. As shown in fig. 4, the deep packet inspection device 21 may include an inspection function entity 41 and an inspection management entity 42, which will be described separately below.

The detection function entity 41 is the body that performs the DPI detection function and may include a memory 43 and a processor 45, and the processor 45 may execute the DPI detection process by executing instructions stored in the memory 43, so the processor 45 may also be referred to as a DIP engine, which may be implemented by a CPU, such as an ARM CPU. In addition to executable instruction codes, the memory 43 may also store related configuration information of the detection function entity 41, such as the detected service type (i.e. application type), the packet parsing rule, and the like. The processor 45 may perform the following steps by executing instructions in the memory 43 and based on configuration information in the memory 43:

intercepting network data packets from one or more underlying networks 10;

parsing the intercepted network data packet to determine a service type and a related object of the network data packet; and

and repackaging the network data packet into a preset format based on the parsing result, and sending the network data packet to the network intelligent processing device.

The format of the network packets may vary from network to network and from application to application. Therefore, in addition to intercepting data, the detection functional entity 41 may parse the data packet according to the configured parsing rule, identify useful information therein, repackage and encapsulate the network data packet according to a predetermined format, and send the network data packet to the network intelligent processing device. By uniformly packaging and encapsulating the network data packets by the detection functional entity 41, a format convenient for the network intelligent processing device to use can be formed, thereby reducing the data processing load at the network intelligent processing device.

The detection management entity 42 may manage the operation of the detection function entity 41. For example, the detection management entity 42 may configure the service types that the detection function entity 41 can detect, the parsing rule for the detection function entity 41 to parse the network data packet, and the like, and the parsing rule may be represented by a regular expression. Although detection function 41 and detection management entity 42 are shown in fig. 4 as two separate modules, in some embodiments, detection management entity 42 may also be implemented in detection function 41 as one management module.

The reconfiguration of the network is not only embodied in the overall architecture described above, but may also be embodied in the base network 12. Software functions are separated and decoupled from a hardware platform, and a basic network is driven to an SDN; the software functions are clouded and open source, and the embedded software deployment operation automation is not needed any more; moving from office-centric networking to cloud Data Center (DC) -centric networking; the architecture is unified, the network capability is opened, the application is flexibly called, and the centralized management and control are realized; -defining services by the operator towards defining services by the user; the traditional supplier trading relation is shifted to the construction of an industrial chain ecosystem. After the network SDN/NFV is formed, hardware management is separated from service operation, the network and the service operation are converted into cloud, networking of physical equipment is changed into virtual software operation, an integrated cloud platform becomes a basic platform of a future network architecture, and then the cloud platform is gradually upgraded to an intelligent architecture. Two logic architectures are required to be built on a network entity, an intelligent architecture and a traditional architecture are built, and the final aim is to build a network intelligence brain and a nervous system. The newly-built intelligent architecture is superposed with a nerve layer, and a nerve system is a component of a super agent or a city agent. The SDN/NFV is the first step to reconstruct the traditional architecture, i.e. to reconstruct the network body, but to ensure the interworking and consistency with the existing network. The two logical architectures have different respective orientations, but with deep intersections and merges between them.

Taking an operator optical network as an example, the software definition architecture is realized as follows: the NFV is used for realizing virtualization and modularization of service and network functions, defining of equipment function software, separating software and hardware, separating control and forwarding, realizing flexible architecture, centralized service control and network slicing, realizing customization of networks and equipment architectures according to needs, and changing the problems of single architecture, inflexibility and rigid structure in the past.

And realizing SDN: soft definition and dynamic connection of a network are realized by using an SDN; the system has the advantages that the system is centered on a user, functions are modularized, resources are fully shared, the capacity is fully opened, strategies are easy to arrange, services are tightly coupled, services are easy to experience, the network can be self-healed, and management can be unified; the intelligent, flexible, open, expandable, efficient, large-capacity and on-demand customization of the network is realized, and the problems of the prior network structure rigidity, complex mode and low operation efficiency are solved.

And (3) realizing software defined operation: services are separated from a network, users are separated from the network, and the SDN and the NFV are used for realizing the on-demand management of the users, performance, flow, configuration, charging, safety and the like; the method has the advantages that the network function and the service are open, the whole process is centralized and managed by software, the service is quickly on-line and unified in operation and maintenance, the service and the strategy are customized according to needs, the network is served according to needs, big data is intelligent, an industrial chain is open, and the traditional single pipeline and solidified operation mode is changed.

Figure 5 illustrates an architectural block diagram of a Software Defined Network (SDN) that may be used to implement the underlying network 12 in an embodiment of the invention, according to an exemplary embodiment of the invention. As shown in fig. 5, the SDN may include an infrastructure layer 52, a control layer 54, and a collaborative orchestration layer 56, each described below.

Infrastructure layer 52 may include various network hardware devices, such as those described above with reference to FIG. 2, as an underlying portion of the separation of network resource software and hardware. Infrastructure layer 52 may utilize the OpenFlow protocol to program data flows based on a flow table (FlowTable) and achieve consistency of flow table updates.

The control layer 54 includes a plurality of controllers for controlling the infrastructure layer 52, such as issuing flow tables and the like. It should be understood that the individual controllers are software-defined controllers, such that the underlying network 10 implements network function virtualization. The control layer 54 centrally controls network resources, so that flexible scheduling and cooperative control of network resources can be achieved.

Collaborative orchestration layer 56 is responsible for collaborative orchestrating network data and is provided to the application layer (not shown) through an open port (API). According to the business requirements, the cooperative arrangement layer 56 can dynamically provide an arrangement template and a control strategy, so as to realize end-to-end arrangement of the data stream.

It can be understood that the SDN network is a cross-layer optimization-oriented OpenFlow-based uniform resource model, and can implement optimization control on a multilayer multi-domain multi-constraint base network. In the embodiment of the invention, through the synergistic effect of the SDN basic network and the upper network agent, more functions of the network can be mined and exerted globally and efficiently, the value of the network is improved, and the rapid development of the productivity is promoted.

In the face of major technical challenges such as network expandability, safety, user experience guarantee, real-time performance, mobility and manageability, and major technical challenges such as pulling economic challenges, manufacturing transformation challenges, big data challenges and other major strategic challenges faced by networks, the application jumps out of a pure network innovation thinking, comprehensively solves the problems existing in the network through an end-to-end and whole-process whole-network-angle 'network (cloud network) + perception + big data + algorithm processing + artificial intelligence' human-like behavior consciousness and a self-regulation method, mutually coordinates and jointly solves the problems existing in the network (because the comprehensive problem is difficult to solve by the pure network alone), and considers compatibility with the traditional network technology and protects the generated network investment. Many of the problems that were not previously well solved in the pure network context may become relatively simple due to the introduction of artificial intelligence and large data analytics. For example, end-to-end user experience guarantees can be coordinated and solved through an end-to-end intelligent bandwidth management policy. For another example, the scalability problem can be cooperatively solved by a wider end-to-end network, and the problem location is only more accurate and at least can be alleviated. Solving the problem of people is a strong item of intelligence, because the source and root of the problem in future networks are also people.

The value of the network is proportional to the square of the number of nodes N in the network, with the greater N the higher the value of the network. The network is the core resource of the country and the city, the network architecture is the soul of the network, and the network revolution is a strategic measure, which has great significance. Network reconfiguration and introduction of intelligence are inevitable and cannot be avoided, and SDN/NFV is the first step and is still in the early development stage at present. Network reconfiguration requires two logical architectures to be built on one entity: the method comprises the steps of establishing an intelligent architecture (such as an urban intelligent agent, and superposing a neural layer on a traditional network architecture) and reconstructing the traditional architecture (a network body, and ensuring the intercommunication and consistency with the current network). The ultimate goal is to build NAIs and network brains.

Only some embodiments of the invention have been described above. It will be understood that various changes in form and details may be made therein without departing from the spirit and scope of the invention, and that such changes are encompassed within the invention as defined by the appended claims and their equivalents.

Claims (21)

1. A network intelligence system, comprising:

the network data sensing device is arranged in one or more basic networks and used for sensing network data in the one or more basic networks; and

and the network intelligent processing device is used for carrying out artificial intelligent processing on the network data obtained from the network data perception device so as to obtain the information of the objects related to the one or more basic networks.

2. The network intelligence system of claim 1, wherein the network data awareness means comprises awareness means for awareness of human society, awareness means for awareness of the natural world, and awareness means for awareness of cyberspace.

3. The network intelligence system of claim 2, wherein the means for perceiving network space comprises at least one deep packet inspection device disposed in the one or more base networks, the deep packet inspection device configured to intercept network data packets from the one or more base networks.

4. A network intelligent system according to claim 3, wherein said deep packet inspection device comprises an inspection function, said inspection function having at least one processor and at least one memory, said memory having instructions stored thereon, said processor executing said instructions to perform the steps of:

intercepting network data packets from the one or more underlying networks;

parsing the intercepted network data packet to determine a service type and a related object of the network data packet; and

repackaging the network data packet into a preset format based on the analysis result and sending the network data packet to the network intelligent processing device.

5. The network intelligence system of claim 4, wherein the deep packet inspection device further comprises an inspection management entity configured to manage operation of the inspection function entity according to instructions of the network intelligence processing device.

6. The network intelligence system of claim 2, wherein the perceiving means for perceiving human society and the perceiving means for perceiving the natural world each comprise:

an automatic sensing unit for automatically sensing the related data; and

and the perception control unit is used for controlling the operation of the automatic perception unit according to the instruction of the network intelligent processing device.

7. A network intelligence system in accordance with claim 1 wherein said artificial intelligence process comprises determining a behavior and/or state of a related object based on said network data.

8. The network intelligence system of claim 7, wherein the artificial intelligence process further comprises determining attributes of the related objects based on their behavior and/or status.

9. The network intelligence system of claim 8, wherein the artificial intelligence process further comprises controlling behavior of the related objects based on attributes of the related objects.

10. The network intelligence system of claim 7, wherein the artificial intelligence process further comprises predicting future behavior and/or state of related objects through machine learning based on the network data.

11. The network intelligence system of claim 10 wherein the network intelligence processing arrangement is configured to run a neural network model that is trained using the obtained network data to predict future behavior and/or state of the relevant object.

12. The network intelligence system of claim 7, wherein the network intelligence processing apparatus is further configured to determine associations between behaviors and/or states of a plurality of related objects through a plurality of network data related to the plurality of objects.

13. The network intelligence system of claim 7, wherein the network intelligence processing device is configured to perform one or more of the following:

determining travel information for the related objects based on the camera data, vehicle ticketing data, accommodation data;

determining traffic load information for the relevant road based on the vehicle position and speed data, and adjusting the traffic control signal and/or planning the navigation path based on the traffic load information;

determining bandwidth demand information for the relevant objects based on the traffic data, and configuring bandwidth and/or planning routes based on the bandwidth demand information;

determining preferences for related objects based on the web browsing data and determining push content for the related objects based on the preferences for the related objects; and

the method further includes determining air haze data based on the haze sensor, and determining a source of haze pollution based on the air haze data.

14. The network intelligence system of claim 1, wherein the one or more base networks comprise a telecommunications network, the internet, and/or the internet of things.

15. The network intelligence system of claim 1, wherein the one or more base networks include one or more of an enterprise network, a community network, a traffic control network, an energy control network, an environmental monitoring network, a government affairs network, a logistics network, a banking network, a school network, a hospital network, a port network, a building network, a metropolitan network, and a backbone network.

16. A network intelligence system in accordance with claim 1 wherein each base network comprises a client, a server, and an intermediate connection for connecting the client and the server.

17. The network intelligence system of claim 16, wherein the client comprises one or more of a personal computer, a cell phone, a tablet computer, a personal digital assistant, and a sensor,

the server comprises one or more of a cloud server, a base station, a server computer and a mainframe, and

the intermediate connections include one or more of repeaters, bridges, routers, switches, and gateways.

18. The network intelligence system of claim 1, wherein the one or more base networks comprise a Software Defined Network (SDN).

19. The network intelligence system of claim 18, wherein the software defined network comprises:

an infrastructure layer comprising a network infrastructure;

a control layer comprising a controller for controlling the network infrastructure, wherein the controller is defined by software; and

and the cooperative arrangement layer is used for performing cooperative arrangement on the network data and providing the network data to the application layer through the open port.

20. The network intelligence system of claim 19, wherein the network intelligence processing arrangement controls operation of an infrastructure layer of the one or more infrastructure networks by configuring a control layer of the one or more infrastructure networks.

21. The network intelligence system of claim 1, wherein the network intelligence processing devices include one or more of enterprise intelligence processing devices, community intelligence processing devices, traffic intelligence processing devices, energy intelligence processing devices, environmental intelligence processing devices, government intelligence processing devices, logistics intelligence processing devices, bank intelligence processing devices, school intelligence processing devices, hospital intelligence processing devices, port intelligence processing devices, building intelligence processing devices, metropolitan area network intelligence processing devices, and backbone network intelligence processing devices, and super intelligence processing devices.

Images