阅读说明：本技术 一种用于基站的深度数据处理的装置及方法 (Device and method for deep data processing of base station ) 是由 季一峰 于 2021-09-10 设计创作，主要内容包括：本发明公开一种用于基站的深度数据处理的装置及方法,属于电通信的技术领域。本发明在基站和UPF/SGW//SGW/核心网之间部署深度学习和处理信令控制报文、IMS协议报文和数据报文的深度数据处理装置,保持移动通信的移动性管理能力和IMS通话能力,通过管理面组件在控制面组件中设置各接口传递数据报文的格式并预设用于加载透传和本地转发处理信息的全局配置表、分配表,控制面组件深度学习报文并生成转发表,用转发面组件实现GTP报文在基站侧做本地处理和极低延时的转发,在有限基站频谱带宽内实现接入更多移动终端,并以更低的延时接入到本地网络或基站运营商网络,大幅提升移动终端的本地访问质量,大幅提升基站频谱利用率,推进移动网络达到碳中和及碳达峰指标。(The invention discloses a device and a method for deep data processing of a base station, and belongs to the technical field of electric communication. The invention arranges a deep data processing device for deep learning and processing signaling control messages, IMS protocol messages and data messages between a base station and a UPF/SGW/core network, keeps the mobility management capability and IMS conversation capability of mobile communication, sets the format of each interface transmission data message in a control surface component through a management surface component, presets a global configuration table and a distribution table for loading transparent transmission and local forwarding processing information, deeply learns the messages and generates a forwarding table by the control surface component, realizes the local processing and extremely low-delay forwarding of GTP messages on the base station side by using the forwarding surface component, realizes the access of more mobile terminals in a limited base station spectrum bandwidth, and accesses to a local network or a base station operator network with lower delay, greatly improves the local access quality of the mobile terminals, and greatly improves the utilization rate of the base station spectrum, the mobile network is propelled to reach the indexes of carbon neutralization and carbon peak reaching.)

1. An apparatus for deep data processing for a base station, comprising:

the control surface component is used for learning GTPU data packets transmitted between a base station and an UPF/SGW/core network and then generating a forwarding table, when the GTPU data packets are signaling messages and IMS protocol messages, the forwarding table is refreshed after the signaling messages are learned, when the GTPU data packets are data messages and meet local processing conditions of the data messages, the forwarding table is refreshed when the GTPU data packets are the data messages but do not meet the local processing conditions of the data messages, the forwarding table is refreshed after the periodic detection message response packets fed back by a learning N32 interface, the ARP packets of the MAC of an LB interface IP requested by a local exit router are learned, the ARP response packets are constructed based on the MAC addresses of the LB interface IP, and the IP data packets sent by the local exit router are learned;

the forwarding plane component inquires the global configuration table and the forwarding table, and locally forwards or transparently transmits the received data packet according to an inquiry result, the GTPU data packet is a signaling message, an IMS protocol message, and the GTPU data packet is transparently transmitted between an N31 interface and an N32 interface when the forwarding table is incomplete, the GTPU data packet is a data message, and is forwarded to an LB interface after a GTP message header is stripped when the GTPU data packet meets the local processing condition of the data message, the GTPU data packet is a data message, and is transparently transmitted between N31 and N32 when the GTPU data packet does not meet the local processing condition of the data message, the ARP response packet generated by the forwarding plane component is forwarded to the LB interface, and is locally forwarded to the N31 interface after the IP data packet is subjected to GTPU message encapsulation;

the management surface component updates or configures a distribution table and a global configuration table preset by the control surface component and is used for inquiring the working state of the device; and a process for the preparation of a coating,

an N31 interface communicated with the base station transmits an uplink GTPU data packet of the base station to the control plane assembly and the forwarding plane assembly, and transmits an IP data packet locally forwarded by the forwarding plane assembly to the base station;

an N32 interface communicated with the UPF/SGW transmits a downlink GTPU data packet of the UPF/SGW/core network to a control plane component and a forwarding plane component, and transmits a signaling message/IMS protocol message or a GTPU data packet transmitted by the forwarding plane component to the UPF/SGW/core network;

the LB interface is communicated with the local exit router, transmits the IP data packet sent by the local exit router to the control surface component and the forwarding surface component, and transmits the data packet or the ARP response packet which is forwarded by the forwarding surface component and is stripped of the GTP message header to the local exit router;

and the M interface for network management transmits the update or configuration information of the distribution table and the global configuration table to the management surface component.

2. The apparatus of claim 1, wherein the allocation tables comprise a local processing user address field allocation table and a server address field allocation table, the global configuration table stores enabling switch status information, a MAC address of a local egress router, an LB interface IP, an upstream port, a downstream port, and a local processing port, and the forwarding table stores: the method comprises the steps of user terminal message IPv4/IPv6 address, address type, base station address, base station MAC address, base station side full tunnel endpoint identification, packaging mark, UPF/SGW address, virtual local area network address, timestamp and local processing state mark.

3. The apparatus of claim 1, wherein the local processing condition of the data packet is: inquiring the IPv4/IPv6 address of the user terminal message in a local processing user address field distribution table, or inquiring the IPv4/IPv6 address of the user terminal message in a server address field distribution table; and writing the IP address of the local data center or the server into a server address field distribution table through the M interface, or writing the source IP address of the mobile terminal into a local processing user address field distribution table through the M interface, or writing the destination IP address of the mobile terminal into the server address field distribution table through the M interface.

4. The apparatus of claim 1, wherein the GTPU data packet comprises: an outer ethernet data packet, an outer IPv4/IPv6 data packet, a UDP data packet, a GTPU data packet, an inner IPv4/IPv6 data packet, and an inner UDP data packet encapsulating an original data packet, where the GTPU signaling packet includes: IPv4/IPv6 data packets, UDP data packets, GTPU data packets.

5. The apparatus of claim 1, wherein the apparatus is further configured to carry N1/S1 protocol for signaling between UE and AMF/MME, N2 protocol for signaling between base station and AMF, N3 protocol for signaling between RAN and UPF/SGW, and IMS service protocol for carrying VoLTE service between UE and IMS.

6. The apparatus of claim 5, wherein the messages transmitted by the S1 protocol, the N1 protocol, the N2 protocol, and the N31 interface and the N32 interface support 802.1Q encapsulation, and the GTPU packets transmitted by each interface include a VLAN TAG for identifying the VLAN to which each protocol or interface belongs.

7. The apparatus of claim 1, wherein the N31 interface and the N32 interface simultaneously support 4G network, 5G network, and VoLTE voice services, and receive GTPU data packets of 4G network and GTPU data packets of 5G network and VoLTE voice service packets.

8. The apparatus of claim 7, wherein the GTPU data packet of the 4G network comprises: flag bit, data packet type, length, and full tunnel endpoint identifier, where the GTPU data packet of the 5G network includes: the method comprises the steps of zone bits, data message types, lengths, full-quantity tunnel endpoint identifications, PDU session container extension information and IMS messages of VoLTE services.

9. The apparatus of claim 1, further comprising an API and a network management interface for querying configuration tables, status tables, forwarding tables, traffic, bandwidth, QoS levels, SLA parameters preset by the apparatus.

10. The apparatus of claim 1, wherein during the learning of GTPU packets transmitted between the base station and the UPF/SGW/core network, the control plane component refreshes the forwarding table once learning is interrupted by the control protocol/IMS protocol, and after querying the forwarding table, the forwarding plane component forwards all traffic of the base station to the LB interface, which is in communication with the local data center core network/local IMS.

11. A method for deep processing of GTPU data packet on base station side, characterized in that, the deep data processing device of any one of claims 1 to 10 is deployed between the base station and UPF/SGW/core network,

when receiving an uplink GTPU data packet from a base station, learning the uplink GTPU data packet and transmitting the uplink GTPU data packet to a UPF/SGW/core network, and updating a TEID, a timestamp and a routing table field in a forwarding table when learning that the base station sends an alarm message;

and transmitting the uplink GTPU data packet to the UPF/SGW/core network or transmitting the uplink GTPU data packet to a local egress router according to the forwarding table.

12. The method of claim 11, wherein the deep learning of the GTPU packet is performed while learning a DNS request sent by a user terminal, and after the DNS request is resolved, the DNS request is locally processed or locally forwarded or transparently passed to the UPF/SGW/core network.

13. The method of claim 11, wherein the method for deep processing of GTPU packets on the base station side learns the IMS protocol packet of VoLTE service between the user equipment and the IMS while deep learning of GTPU packets, refreshes the forwarding table, and directly passes through the IMS protocol packet to the IMS network element.

14. A method for deeply processing downlink GTPU data packets on UPF/SGW/core network side, characterized in that the deep data processing device of any one of claims 1 to 10 is deployed between a base station and UPF/SGW/core network,

when receiving a downlink GTPU data packet from a UPF/SGW/core network, learning the downlink GTPU data packet and then transmitting the downlink GTPU data packet to a base station;

deeply learning GTPU messages transmitted between a base station and a UPF/SGW/core network, and updating a base station side full tunnel endpoint identification field, a user terminal message IPv4/IPv6 address and a virtual local area network address in a forwarding table; when the N32 interface feeds back the periodic detection message response packet, the forwarding table is updated after the periodic detection message response packet is learned;

and deeply learning the destination IP of the downlink GTPU data packet, updating a field related to the base station in a forwarding table, and transparently transmitting the downlink data packet to the base station.

Technical Field

The invention discloses a Deep Data Processing Unit (DDPU) and a method for a base station, relates to a mobile communication network technology, and belongs to the technical field of electric communication.

Background

The multi-access edge computing and mobile network local data center transfers intensive computing tasks to a mobile network edge server near a base station, so that the congestion and the burden of a mobile core network and a mobile transmission network are reduced, the data are computed and processed at a near end, the bandwidth pressure of a mobile bearing network is relieved, the low time delay and the large bandwidth are realized, the data processing efficiency in the universal internet era is improved, the data request of a mobile terminal user can be quickly responded at the base station side, and the service quality is obviously improved.

The GTP Protocol (GPRS tunneling Protocol) provides Protocol channels between GSNs in the GPRS backbone network (e.g., SGSN and GGSN), and PDUs of all packet data protocols should be encapsulated by the GTP Protocol. How to fast forward the GTP message at the base station side is a key problem to increase the number of mobile terminal accesses, and due to the limitation of the von neumann architecture, the server device has a difference of 1000 times between the forwarding and processing speeds of the message and the actual demand, and cannot meet the characteristic requirement of low latency of the network. The switch and the conventional two-layer or three-layer routing switching equipment such as the hub, the shunt, the TAP and the like do not have the capacity of deep processing and classification of messages, the analysis and the processing of 4G/5G mobile communication protocol signaling messages at a base station side, the analysis and the processing of IMS protocol signaling messages and the functions of simultaneously realizing local forwarding and local low-delay communication after deep processing of data messages at the base station side. In addition, the traditional shunt and TAP equipment forcibly shunt the signaling message/IMS protocol message, which results in the loss of control signaling, inevitably causes the mobile terminal to drop the line and call, and affects the network access attachment of the mobile terminal; some technical solutions for increasing the message forwarding and processing speed require reconfiguration of the base station, the terminal, the core network, the UPF/SGW, and the mobile bearer network, and this increases the complexity of implementation undoubtedly by reconfiguring the breakout message at the UPF/SGW or core network side. The method aims to provide a technical scheme for realizing and accelerating the forwarding and processing of the data flow of the mobile terminal at the local extremely low delay of the base station side on the premise of not influencing the mobile terminal service (surfing and making a call) and zero configuration of a base station, a terminal, a core network, a UPF/SGW and a mobile bearer network.

Disclosure of Invention

The present invention is directed to overcoming the above-mentioned shortcomings in the art, and provides an apparatus and method for deep data processing of a base station, on the premise of not influencing the network access attachment, the internet surfing, the telephone calling and the zero configuration of a base station, a terminal, a core network, a UPF/SGW and a mobile bearer network of a mobile terminal, the technical problem that the existing two-layer or three-layer forwarding equipment, SDN equipment, TAP equipment, a server and the like are not suitable for processing a wireless communication network scene of 4G/5G protocol messages through a GTP protocol is solved, the frequency spectrum utilization rate of the base station is maximized, the invention can access more user terminals and simultaneously realize the purpose of ultralow delay, and promotes the realization of 4G/5G/6G edge calculation application scenes with huge link, large bandwidth and extremely low delay of industrial and civil mobile internet from the aspect of deep learning and processing of data entering and leaving a base station.

The invention adopts the following technical scheme for realizing the aim of the invention:

an apparatus for deep data processing for a base station, comprising: an N31 interface for communicating with a base station, an N32 interface for communicating with a UPF/SGW, an LB interface for communicating with a local egress router, an M interface for network management, a control plane component, a forwarding plane component, a management plane component; the control surface component receives data input by an N31 interface, an N32 interface and an LB interface, a forwarding table is generated after deep learning of a message transmitted between a base station and a UPF/SGW/core network, and meanwhile, a global configuration table and a distribution table which are preset by the control surface and used for loading transparent transmission and local processing information are updated through the management surface component; and the forwarding plane component transparently transmits or locally forwards the data input by the N31 interface, the N32 interface and the LB interface by inquiring the global configuration table and the forwarding table.

And analyzing a GTP data packet transmitted between the base station and the UPF/SGW/core network, and when the data packet is a signaling message/IMS protocol message, the DDPU learns the N1/N2/N3/S1/IMS protocol message and refreshes a forwarding table of the DDPU to perform transparent transmission processing.

And when the GTPU data packet is a data packet and meets the local processing condition of the data packet, the DDPU strips off a GTP packet head and sends an IP data packet to the local data center and the local outlet router A through the LB interface, and the reverse DDPU sends the IP packet from the LB and the GTP head to the base station to realize the bidirectional direct communication between the mobile terminal and the local data center as well as the outlet router A.

And the DDPU analyzes the GTPU packet sent to the remote router B, and transparently transmits the GTPU packet between the base station and the UPF/SGW/core network when the local processing condition of the data message is not met.

And analyzing the ARP packet sent by the local exit router A, and when the exit router A requests the MAC of the communication interface IP, customizing and sending the ARP constructed based on the MAC address of the communication interface of the local exit router to the exit router by the DDPU.

Further, a device for deep data processing of a base station, a control plane component stores a global configuration table, a local processing user address field allocation table, a server address field allocation table and a forwarding table, the global configuration table stores enabling switch state information, an MAC address of an egress router a, a local egress IP, an uplink port, a downlink port and a offload port, and the forwarding table stores: user terminal message IPv4/IPv6 address, address type, base station address, base station MAC address, base station side full tunnel endpoint identification, packaging mark, UPF/SGW address, virtual local area network address, timestamp and shunting state mark.

Further, a device for deep data processing of a base station, the method for determining that the local processing condition of the data packet is satisfied is: the user terminal message IPv4/IPv6 address can be inquired in the distribution table of the distribution user address field, or whether the user terminal message IPv4/IPv6 address can be inquired in the distribution table of the server address field.

Further, an apparatus for deep data processing of a base station, where a GTPU data packet includes: an outer ethernet data packet, an outer IPv4/IPv6 data packet, a UDP data packet, a GTPU data packet, an inner IPv4/IPv6 data packet, and an inner UDP data packet encapsulating an original data packet, where the GTPU signaling packet includes: IPv4/IPv6 data packets, UDP data packets, GTPU data packets.

Further, an apparatus for deep data processing of a base station, the control plane component and the forwarding plane component simultaneously supporting: an N1/S1 protocol for carrying signaling transfer between the UE and the AMF/MME, an N2 protocol for carrying signaling transfer between the base station and the AMF, an N3/S1 protocol for carrying signaling transfer between the RAN and the UPF/SGW, and an IMS protocol for carrying VoLTE traffic between the UE and the IMS.

Further, a device for deep data processing of a base station, where packets transmitted by an S1 protocol, an N1 protocol, an N2 protocol, an IMS protocol, an N31 interface, and an N32 interface support 802.1Q encapsulation, and GTPU packets transmitted by each protocol and interface all include a VLAN TAG for identifying a VLAN to which each interface belongs.

Further, an N31 interface and an N32 interface of the device for deep data processing of a base station simultaneously support a 4G network and a 5G network, receive GTPU data packets of the 4G network and GTPU data packets of the 5G network, and simultaneously support VoLTE voice service. Meanwhile, the evolution system of subsequent mobile communication can be supported by upgrading the protocol field.

Further, an apparatus for deep data processing of a base station, a GTPU data packet of a 4G network includes: flag bit, data packet type, length, and full tunnel endpoint identifier, where the GTPU data packet of the 5G network includes: zone bit, data message type, length, full tunnel endpoint identification, PDU session container extension information.

Further, a device for deep data processing of a base station, wherein a control plane component realizes a DDPU deep learning engine function, and performs passive learning on a message transmitted between the base station and a UPF/SGW/core network: an alarm message sent by an uplink deep learning base station, a GTPU message between the downlink deep learning base station and a UPF/SGW/core network are learned to obtain the association relation between fields such as TEID/IP/vlan tag/timestamp and a 3GPP protocol, the association relation is recorded into a forwarding table of a DDPU after DPDU (digital pre-distortion unit) identification processing, and the forwarding table of the DDPU is refreshed in a plurality of clock pulse periods so as to maintain the continuous availability of an internet service between a mobile terminal and the base station.

Further, an apparatus for deep data processing of a base station, the DDPU deep learning engine further has an active learning function: the deep learning engine periodically and actively sends a message to an N32 interface, learns after receiving a return packet of UPF/SGW/PGW, maintains the mobility management capability of the mobile terminal and the mobile core network, records the return packet into a forwarding table of the DDPU after DPDU identification processing, and refreshes the forwarding table of the DDPU in a plurality of clock pulse periods so as to maintain the continuous availability of the internet service between the mobile terminal and the base station.

Further, a device for deep data processing of a base station, DDPU can learn a DNS request message sent by a terminal, and because a DNS server of a local egress router a and a DNS server of a core network are inconsistent, according to a service requirement of a CDN, a DNS request is processed or forwarded to an LB port in the device or directly passed through to a remote core network, so as to achieve correct and fast resolution of a domain name of a mobile terminal.

Furthermore, the DDPU externally provides a network management M interface of a software API and hardware for an administrator to configure an allocation table and a configuration table and inquire the flow, the bandwidth, the QoS level and the SLA parameter of the DDPU so that the administrator can use the DDPU for management and charging.

Furthermore, a device for deep data processing of base station, writing in the address field distribution table of the target server to realize the access of the mobile terminal to the local network of the base station side, and for the zero configuration of network architecture and operation and maintenance, only the IP address range of the server or data center is written in the server address field distribution table of the DDPU through the management port of the DDPU to realize the bidirectional communication between the mobile terminal and the local server or data center of the base station side

Further, a device for deep data processing of a base station writes in a source IP section of a mobile terminal to realize accessing to a local network of the base station side, configures network architecture and operation and maintenance zero, and writes an IP address range of the mobile terminal into a distribution user address section distribution table of a DDPU through a management port to realize bidirectional communication between the mobile terminal and a local server or a local data center or a far-end operator network of the base station side. The source IP section range of the mobile terminal is not in the DDPU mobile terminal, and the network is accessed according to the original path planned by the operator and is not controlled by the DDPU.

Furthermore, a device for deep data processing of a base station supports a mobile terminal to access a public network and a local private network simultaneously, configures network architecture and operation and maintenance zero, and writes a target IP address range of the mobile terminal to be accessed into a server address segment distribution table of a DDPU (distributed data Unit) through a management port to realize the simultaneous bidirectional communication of the mobile terminal, the local private network at the base station side and a public bearer network of the base station.

Further, the device for deep data processing of the base station forwards all traffic of the base station to the local data center through the LB port, so that local redundant backup of a core network/IMS is realized, and normal work of internet access service and VoLTE service of a local mobile network is guaranteed in an application scene of inelegance such as fire disaster of an operator network.

After the device is deployed between a base station and a UPF/SGW/core network, the device processes an uplink GTPU data packet transmitted by the base station, a downlink GTPU data packet transmitted by the UPF/SGW/core network and an IP data packet transmitted by a local outlet router.

By adopting the technical scheme, the invention has the following beneficial effects:

(1) the invention realizes the local processing of the data service at the base station side after deeply learning and processing the data message at the base station side, processes the signaling message and the control message through a device arranged between the base station and the UPF/SGW/core network, processes the local data message in real time, sets the message format of each interface for transmitting the data message on a software control surface, presets a global configuration table for loading transparent transmission and local processing information, updates a forwarding table in real time after deeply learning the message transmitted between the base station and the UPF/SGW/core network, updates the configuration table and the distribution table through a network management interface, realizes the acceleration of the data flow of the mobile terminal to be forwarded and processed at the local extremely low delay at the base station side, realizes that more user terminals access the local production network with lower delay in the limited base station spectrum bandwidth, the carbon neutralization carbon peak-reaching index is achieved by improving the frequency spectrum utilization rate of the base station. When the deep data processing device provided by the invention is used for locally processing the base station data service, the user mobile terminal is always on line, the internet access and VoLTE service are not interrupted, and the time delay lower than that required in the 3GPP R16 standard of 5G is achieved.

(2) The invention writes the source IP and the target IP required by the bidirectional communication into the local processing user address field distribution table and the server address field distribution table through the communication management interface, namely, the mobile terminal can access the local network, the public network and the local private network simultaneously on the premise of zero configuration of the network architecture, and compared with the scheme of realizing message distribution by reconfiguring the base station, the terminal, the core network, the UPF/SGW and the mobile carrier network, the invention has the technical advantages of simpleness and easy realization.

Drawings

Fig. 1 is a network architecture for implementing deep data processing of a base station according to the present invention.

Fig. 2 is a schematic diagram of forwarding after DDPU locally processes 4IN4 tunnel messages.

Fig. 3 is a schematic diagram of forwarding after DDPU locally processes 4IN6 tunnel messages.

Fig. 4 shows a typical networking architecture of 4/5G with simultaneous access to DDPU.

Fig. 5 is a schematic diagram of the DDPU supporting 802.1Q encapsulation messaging.

FIG. 6 is a block diagram of DDPU functional modules.

Fig. 7 is a flowchart of DDPU processing an upstream packet at the N31 interface.

Fig. 8 is a flowchart of DDPU processing downstream packets of the N32 interface.

Fig. 9 is a flowchart of processing an LB interface downlink packet by a DDPU.

Detailed Description

The technical scheme of the invention is explained in detail in the following with reference to the attached drawings.

As shown in fig. 1, the present invention uses a deep data processing device DDPU between a base station (e.g., a 5G base station gNB) and a User Plane Function (UPF)/SGW to implement GTP packet local processing and very low latency forwarding. The DDPU: the N1/S1 protocol for carrying signaling transfer between the UE and the AMF/MME, the N2 protocol for carrying signaling transfer between the base station and the AMF, the N3/S1 protocol for carrying signaling transfer between the RAN and the UPF/SGW, and the IMS protocol for carrying VoLTE service between the UE and the IMS comprise: an N31 interface for interfacing with a base station, an N32 interface for interfacing with a UPF/SGW, an LB interface for interfacing with a Local Breakout router a (hereinafter referred to as Local Breakout), and a network management M interface, which are used to locally process data packets of a mobile phone user transmitted between the base station and the UPF/SGW, as shown in fig. 1, and transmit data packets, which originally come in and come out from a telecommunication Breakout router B, of a mobile terminal to a Local data center and a server and to communicate with the Local Breakout router a.

After receiving the upstream GTPU data packet from the N31 interface, the DDPU determines whether Local Breakout is needed, if so, strips off the header of the GTPU data packet, and sends the GTPU payload (i.e., the original data packet sent by the mobile phone) from the Local Breakout interface to the Local egress router a.

For an original data packet between a base station and a UPF/SGW, two ways of judging whether a GTPU data message needs local processing are provided, one is according to whether an IP address of a terminal user is in a configured terminal IP address field, and the other is according to whether a server IP address accessed by a terminal is in a configured server address field configuration table. If the server address field configuration table is empty, it indicates that all data messages of the mobile terminal user need to be processed locally.

For GTPU signaling messages between a base station and UPF/SGW, the DDPU only performs deep learning and transparent transmission processing on the GTPU signaling messages/IMS protocol messages and does not perform local forwarding processing. For example, after receiving the uplink GTPU data packet from the N31 interface, the uplink GTPU data packet is determined to be a signaling packet and an IMS protocol packet, and then is transmitted from the N32 interface two-layer to the UPF/SGW; and after receiving the downlink GTPU data message from the N32 interface, judging the downlink GTPU data message as a GTPU signaling message and an IMS protocol message, and transmitting the downlink GTPU data message to the gNB base station from the N31 interface two-layer.

In the mobility process of the UE, during a gNB handover or a radio signal strength change process, an FTEID (Full-scale Tunnel Endpoint Identification) on the side of the gNB base station is released when the UE is idle, and the FTEID is reallocated when the UE and the gNB recover a connected state, so that a change of the FTEID needs to be tracked. And the DDPU deeply learns the GTPU message between the base station and the UPF/SGW/core network at regular intervals.

As shown in fig. 6, the DDPU is composed of three parts, a control plane and a forwarding plane, and a management plane. The control plane learns and negotiates communication by formulating a 4G/5G protocol message, designs a format for transmitting a data message by each interface and presets a global configuration table for loading address information of each interface and a forwarding table for loading local processing information at the control plane, inquires the preset configuration table and the forwarding table after receiving service data of a base station or a UPF/SGW/core network, learns a data packet between the base station and the UPF/SGW/core network in real time and refreshes the forwarding table; after inquiring the global configuration table and the forwarding table, the forwarding plane locally forwards or transparently transmits data to a corresponding interface, namely an IMS protocol message, an N1/S1 protocol, an N2 protocol, an N3 protocol, an N31 interface, an N32 interface, an LB interface transparent transmission message or a wireless terminal user data message of a local processing base station; the management surface responds to the control instruction input by the network management M interface to update the global configuration table and the distribution table, and an administrator can inquire the state of the DDPU. The control plane, the forwarding plane and the management plane are realized by software, or the hardware circuit of each interface can be realized by FPGA and ASIC chip technology, and the functions of the control plane and the forwarding plane can be realized by logic control.

The configuration table comprises a global configuration table, a local processing user address field distribution table and a server address field distribution table. The global configuration table is shown in table 1, and records enable SWITCH state information SWITCH, the MAC address of the egress router a, a local processing egress IP, an uplink port, a downlink port, and a local processing port; the forwarding user address field allocation table is shown in table 2; the server address field allocation table is shown in table 3; the forwarding table records user terminal message IPv4/IPv6 addresses UE-IPv4/IPv6-ADDR (key), address Type ADDR-Type, base station address GNB-ADDR, base station MAC address GNB-MAC, base station side full tunnel endpoint identification GNB-TEID, encapsulation mark QFI, UPF/SGW address UPF/SGW-ADDR, virtual local area network address VLAN-ID, TimeStamp and local processing STATE mark STATE as shown in Table 4. QFI of 000 means that the base station currently accessed to DDPU is a 4G base station, and at this time, QFI information needs to be packaged into a GTPU data packet; QFI of 001 indicates that the base station currently accessed to DDPU is a 5G base station, and at this time, no QFI information needs to be packaged into a GTPU data packet. The smoothing supports a mobile communication system of subsequent evolution, such as 6G.

TABLE 1 Global configuration Table

Table 2 local processing subscriber address field allocation table

Server address field
	1.2.3.4-2.3.4.5
202.1.1.1-202.1.1.100

Table 3 server address field allocation table

Table 4 forwarding table

The deep data processing device disclosed by the invention realizes the following functions through a control plane, a forwarding plane and a management plane: (1) enabling a switch to turn on or off the DDPU; (2) inquiring user information, and inquiring the locally processed user information on the DDPU by using a command, such as an IP address of a user, the starting time of local processing, an FTEID at a base station side and an FTEID at a UPF/SGW side; (3) local processing is carried out as required, whether the local processing is needed or not is judged according to the IP address of the user or the IP address accessed by the user, namely, whether the IP address of the user is covered by a user address field or whether the IP address accessed by the user is covered by a server address field is inquired, and if the IP address is covered, the local processing is needed; (4) the v4inv4 data packet is processed locally, the IPv4 user message encapsulated by the IPv4 tunnel received from the base station side can be processed locally, and the downlink message received from the local exit router A can be encapsulated by the GTPU tunnel and sent to the base station; (5) the v4inv6 data packet is processed locally, the IPv4 user message encapsulated by the IPv6 tunnel received from the base station side can be transmitted to the local exit router A after local processing, and the downlink message received from the local exit router A can be encapsulated by the GTPU tunnel and sent to the base station; (6) supporting 4/5G base stations or a mobile communication system of future evolution, and simultaneously supporting 4/5G/6G/VoLTE user access and local processing of messages; (7) the method supports 802.1Q encapsulation and local processing of 802.1Q encapsulation messages on an N3 interface, and because the N3 interface actually shares physical connection with an S1/N1/N2 protocol/IMS protocol, the S1/N1/N2 protocol/IMS protocol also adopts 802.1Q encapsulation, and S1/N1/N2/IMS protocol messages can be forwarded normally; (8) and aging the forwarding table, wherein the forwarding table which has no message for a long time needs to be aged. (9) Processing a DNS protocol message, wherein a DDPU can learn a DNS request message sent by a terminal, and because the DNS server of a local exit router A is inconsistent with the DNS server of a core network, the DNS request is processed or forwarded to an LB port in the equipment or directly transmitted to the remote core network according to the service requirement of the CDN, so that the domain name of the mobile terminal is correctly and quickly analyzed; (10) network management, DDPU provides the network management M interface of software API and hardware for the administrator to configure the distribution list and the configuration list and inquire the flow, bandwidth, QoS level and SLA parameter of DDPU, so as to facilitate the user to use for management and charging; (11) and through the processing of the IMS protocol message, the DDPU can learn the VoLTE service message between the terminal and the IMS network element, so that the VoLTE service is directly transmitted to the remote IMS network element in the equipment, and the VoLTE calling service of the mobile terminal is not interrupted. (12) And when the main core network/IMS disaster recovery and the remote main core network and the IMS have the problem of inefficacy, the DDPU can switch all the traffic of the base station to the local core network and the IMS, thereby ensuring that the mobile production network is always available.

The DDPU forwarding plane queries the global configuration table after receiving the user message to obtain the state of the enabling switch, the enabling switch defaults to be in a closed state, namely the DDPU only deeply learns and transparently transmits the signaling message and the IMS protocol message without locally processing the original data packet, and when the DDPU forwarding plane queries that the enabling switch is in an open state, the DDPU locally processes and forwards the user data message according to the forwarding table transparently transmitted signaling message and the IMS protocol message updated by the deep learning function.

DDPU supports the configuration of a destination server list accessed by the UE, the configuration of a single address and the configuration with a mask. The DDPU forwarding plane checks the server address field configuration table when receiving the user message and the enabling switch is in the open state, if the IP address accessed by the UE is in the coverage range of the server address field configuration table, all the uplink user data messages need to be forwarded to the local outlet router A after being locally processed, otherwise, the uplink user data messages do not need to be locally processed and are directly transmitted to the N32 interface. And if the server address field configuration table is empty and the enabling switch is turned on, the uplink user data message is forwarded after local processing.

When the user message received by the DDPU forwarding plane is inquired to be locally processed, the GNB-ADDR, the UPF/SGW-ADDR and the GNB-TEID in the forwarding plane are inquired, and the message is locally processed according to the GTPU tunnel address type, the base station address and the user terminal address.

The GTPU tunnel is IPv4, that is, the address of the base station and the address of the UPF/SGW are both IPv4 addresses, when the source IP address and the destination IP address IN the user message are IPv4 addresses, the local processing process of the 4IN4 tunnel message is as shown IN fig. 2, after a V4IN V4 packet with IPv4 address 192.168.100.100 received by the base station is transmitted to the DDPU, the DDPU learns the packet and transparently transmits the signaling message to the UPF/SGW with IPv4 address, and the V4IN V4 packet forwarded to the local interface from the DDPU needs to be stripped of the outer layer encapsulation and then transmitted to the egress router a with IPv4 address 192.168.100.101.

The GTPU tunnel is IPv6, that is, the address of the base station and the address of the UPF/SGW are both IPv6 addresses, and when the source IP address and the destination IP address IN the user packet are still IPv4 addresses, the local processing process of the 4IN6 tunnel packet is as shown IN fig. 3, because the local interface is always an IPv4 address, the data packet received by the base station cannot be directly transmitted to the egress router a, and the DDPU needs to allocate a virtual IPv4/IPv6 address IN the same network segment as the egress router to the data packet before forwarding the data packet with the IPv6 address to the egress router a. When the exit router A requests the MAC to the virtual IPv4 Address, the DDPU supports responding to a dynamic ARP (Address Resolution Protocol) request; or, when the egress router a supports configuring the static ARP, the DDPU may not support the static configuring ARP function.

DDPU supports 4/5G base station simultaneous access, and a typical network for 4/5G simultaneous access is shown in FIG. 4. The message types entering from the N31 interface of the DDPU mainly comprise S1/S1-U interface/IMS messages of a 4G network, and N1/N2/N3 interfaces and IMS messages of a 5G network. The S1/N1/N2 messages are all based on SCTP load bearing and belong to signaling messages, and the DDPU only needs to learn and transmit transparently.

The S1-U interface is a user plane interface, which is an interface between a 4G base station and an SGW (Serving GateWay) of a 4G network, and the message of the interface is also GTPU encapsulation, which is different from the GTPU message format of a 5G network in that it has no PDU SESSION contact extension header, and the GTPU message of the 5G network may also contain an IMS message supporting VoLTE service. GTPU packet formats of 4G networks and 5G networks are both packet formats of 3GPP standards, see tables 5 and 6. For the mobile communication system of the future evolution, the extension head of the special data service of the mobile communication system of the future evolution is added in the GTPU message, so that the access of a multi-system communication network can be met.

Flags

Message Type

Length

TEID

Table 54G network GTPU data packet encapsulation format

Table 65G network GTPU data packet encapsulation format

The messages of the N1/N2/N3/S1/S1-U/IMS protocol or interface need to support 802.1Q encapsulation. Under different environments, it is possible that these interfaces belong to different VLANs, so the most flexible implementation is that the DDPU learns and perceives the VLANs to which these interfaces belong from the dynamic depth of the packet.

For S1/N1/N2 signaling messages, GTPU signaling messages and IMS messages, only learning and recording to a forwarding table and then performing transparent transmission processing are needed, and a DDPU can learn and judge whether 802.1Q encapsulation exists or not when learning the signaling messages.

For S1-U/N3 GTPU data packets, there is support to strip packets with 802.1Q encapsulation and send them from LB interface to egress router A. For the downlink original message received from the LB interface, after GTPU encapsulation is performed, when L2 encapsulation is performed on the outer IP message, the VLAN tag to which S1-U/N3 belongs is encapsulated at the same time. A schematic diagram of the DDPU supporting 802.1Q encapsulation messaging is shown in fig. 5.

The encapsulation format of the GTPU data packet is shown in table 7, and the source address and the destination address of the outer IPv4 protocol layer are the address of the UPF/SGW and the base station, respectively. The outer IPv4/IPv6 protocol layer is followed by a UDP protocol layer, the destination port of the UDP protocol layer must be agreed in the 3GPP standard, and the source port is determined by the sending end. The UDP protocol layer is followed by a GTPU protocol layer, typically 8 or 12 bytes for a 4G network and 16 bytes for a 5G network. The Message Type is loaded in the GTPU protocol layer, and when the Message Type is a specific value, the Message Type indicates that the current Message is a data Message, otherwise, the current Message is invalid data or a signaling Message. The GTPU protocol layer is followed by an inner layer IPv4/IPv6 protocol layer, the inner layer IPv4/IPv6 protocol layer is packaged with an inner layer UDP, and the inner layer UDP is packaged with an original data message, namely the original data message transmitted between a user terminal and the Internet.

TABLE 7 encapsulation format of GTPU data packet

The GTPU signaling message format is shown in table 8. The Message Type loaded in the GTPU protocol layer indicates that the signaling Message does not carry user data, only the information is exchanged between the base station and the UPF/SGW, the Message of the Type is learned and transparently transmitted between the base station and the UPF/SGW through the DDPU, and cannot be forwarded to the local outlet router A. The DDPU can obtain the control signaling in real time and transmit the control signaling to a destination terminal in time by deeply learning the signaling message forwarding table, so that the mobile terminal is always attached to a base station and is not disconnected, which is obviously different from the mirror image shunting processing of the traditional shunt and TAP equipment on the signaling message.

TABLE 8

As shown in fig. 6, the deep learning process is as follows: the DDPU deep learning engine deeply learns the alarm message sent by the base station through an N31 interface in the uplink direction, and deeply learns the GTPU message between the base station and the UPF/SGW/core network through an N32 interface in the downlink direction. After learning the association relation between fields such as TEID/IP/VLAN tag/timestamp and 3GPP protocol, the association relation is recorded into a forwarding table of DDPU after DDPU identification processing, and the forwarding table of DDPU is refreshed in a plurality of clock pulse periods so as to maintain continuous availability of internet access service between the mobile terminal and the base station. The forwarding table is realized by TCAM and RAM/DRAM devices to maintain continuous availability of internet access service between the mobile terminal and the base station, and mobility management capability of the mobile terminal and the mobile core network is maintained through the DDPU deep learning engine, otherwise the mobile terminal is disconnected. The deep learning engine periodically and actively sends messages to an N32 interface, learns after receiving a UPF/SGW return packet, records the UPF/SGW return packet into a forwarding table of the DDPU after being identified and processed by the DDPU, and achieves the aim of maintaining the mobility management capability of the mobile terminal and the mobile core network.

When a far-end core network and an IMS (IP multimedia subsystem) have faults, a deep learning engine finds that a control protocol/IMS (IP multimedia subsystem) protocol interruption can not establish a link in the process of learning uplink and downlink messages of a base station, automatically or manually refreshes a forwarding table to be switched to a local core network and the IMS, a forwarding plane component forwards uplink and downlink flows of the base station to an LB (local bus) port, the LB port is connected with the local core network and the local IMS, the mobile communication service of the base station is recovered, the redundancy protection of the mobile communication is realized, and the local production network of the mobile communication is ensured not to be interrupted.

After receiving the uplink data packet from the base station, the N31 interface performs processing according to the flow shown in fig. 7:

the forwarding plane inquires the state of an enabling switch in the global configuration table, learns the uplink data packet when the enabling state switch is in a closed state, updates the forwarding table, transparently transmits the uplink data packet to an N32 interface, and judges whether the uplink data packet is a data message when the enabling state switch is in an open state;

when the uplink data packet is a signaling packet, the control plane detects whether the N31 interface receives an alarm packet sent by the base station, if the alarm packet is received, the TEID, the timestamp and the routing table field in the forwarding table are forcibly updated, otherwise, the signaling packet is learned and then transmitted to the N32 interface, and the uplink data packet is deeply processed when the uplink data packet is a data packet;

reading routing table and time stamp information in a forwarding table, directly transmitting an uplink data packet to an N32 interface, and inquiring a configuration table to judge whether local processing needs to be carried out on the uplink data packet;

inquiring a local processing user address field configuration table or a server address field configuration table to know whether the current data message needs to be locally processed or not, and locally processing the uplink data packet according to the GTPU tunnel address type, the base station address and the user terminal address;

when the uplink data packet is a v4inv4 data packet, transparently transmitting the uplink data packet to an N32 interface when the uplink data packet is an invalid data packet, when the uplink data packet is an effective data packet and receives a DNS request, transparently transmitting the uplink data packet to an N32 interface, when the uplink data packet does not contain the original data packet or the uplink data packet contains the original data packet but does not receive the DNS request, searching a forwarding table by using a source IP address recorded in a GTPU (GTPU) packet, when the uplink data packet is a v4inv6 data packet, locally processing the v4inv6 data packet by the same method as the v4inv4 data packet, allocating a virtual IPv4/IPv6 address to the processed message only before the uplink data packet is forwarded to a local exit router, and forwarding the locally processed message;

and when the forwarding table is successfully searched and the forwarding table information is complete, acquiring the MAC address of the egress router A from the forwarding table, performing L2 encapsulation on the uplink data packet, forwarding the data packet encapsulated by L2 to the egress router A from an LB port, transparently transmitting the uplink data packet to an N32 interface when the forwarding table is incomplete, if the forwarding table is unsuccessfully searched according to the source IP address recorded in the GTPU message, newly establishing the forwarding table according to the source IP address recorded in the GTPU message, setting the STATE to be 0, and transparently transmitting the uplink data packet to the N32 interface.

After receiving the downlink IPv4/IPv6 data packet from the UPF/SGW, the N32 interface performs processing according to the flow shown in fig. 8:

the forwarding plane inquires the state of an enabling switch in the global configuration table, transparently transmits a downlink data packet to an N31 interface when the enabling switch is in a closed state, and judges whether the DDPU is in an active learning state when the enabling switch is in an open state;

if the DDPU is in a passive learning state, namely a GTPU message transmitted between the deep learning base station and the UPF/SGW/core network, updating a timestamp field in the forwarding table, further updating TEID, IP and vlan tag together, and updating the forwarding table in a plurality of clock pulse periods;

in parallel with the passive learning, the deep learning engine periodically sends a message to an N32 interface for active learning, a forwarding table is updated according to a received UPF/SGW/PGW packet in the active learning process, the forwarding table is refreshed in a plurality of clock pulse periods, and the forwarding table updated through the active learning and the passive learning is used for subsequent local forwarding and transparent transmission;

when the downlink data packet is judged to be a signaling message, the downlink data packet is transmitted to an N31 interface in a transparent mode, and local processing is carried out when the downlink data packet is judged to be a data message;

and when the downlink data packet is an invalid data packet, the downlink data packet is transmitted to an N31 interface in a transparent mode, when the forwarding table is successfully searched according to the destination IP address of the inner layer IPv4/IPv6 when the downlink data packet is an valid data packet, the GNB-IPV4-ADDR, the GNB-TEID and the QFI in the forwarding table are updated (the updating is carried out only when the QFI exists in the GTPU Header of the downlink data packet), the STATE is modified to be in a local processing completion STATE, the downlink data packet is transmitted to the N31 interface in a transparent mode, and if the forwarding table is failed to be searched according to the destination IP address of the inner layer IPv4, the downlink data packet is directly transmitted to the N31 interface in a transparent mode.

When the local egress router sends a downlink IP packet, the process is performed according to the flow shown in fig. 9:

because the local outlet router and the base station can not be communicated in two layers, the local router firstly sends an ARP request to the LB interface to obtain the MAC of the IPv4 address of the LB interface, and the DDPU responds to the virtual MAC address after analyzing the ARP request;

after receiving the virtual MAC from the LB interface, the local exit router A queries the state of an enabling switch in a global configuration table by a forwarding plane, discards a downlink IP data packet when the enabling switch is in a closed state, and checks the protocol type of the downlink IP data packet;

searching a destination IP address in a forwarding table according to the protocol type of the downlink IP data packet;

if the destination IP address is successfully inquired, inquiring the STATE of the forwarding table, and if the forwarding table is searched according to the destination IP address and fails, directly transmitting the downlink data to the base station;

and discarding the downlink data packet when the STATE is in a local processing unfinished STATE, and sending the downlink data packet to the base station after GTP, UDP and outer IP encapsulation are carried out on the downlink data packet when the STATE is in a local processing finished STATE.

The GTP encapsulation information comprises a base station side TEID and QFI, QFI does not need to be encapsulated when the 4G base station is accessed into the DDPU, and QFI needs to be encapsulated when the 5G base station is accessed into the DDPU. The outer layer package comprises an L2 layer package, an L3 layer package and an L4 layer package, and the information of the L2 layer package comprises: the destination MAC is the base station MAC, the source MAC is the router MAC, and the information encapsulated by the L3 layer includes: the destination address is the base station address, the source address is the UPF/SGW address, and the information encapsulated by the L4 layer is the source and destination port is 2152.

The mobile terminal can access the local network of the base station side by writing in the address field of the target server, and the IP address of the local data center or the server is written in the address field distribution table of the server through the network management M interface, so that the bidirectional communication between the mobile terminal and the local server or the data center of the base station side can be realized.

The mobile terminal can access the local network of the base station side by writing in the source IP address field of the mobile terminal, and the source IP address of the mobile terminal is written in the local processing user address field distribution table through the network management M interface, so that the bidirectional communication between the mobile terminal and the local server or the local data center of the base station side or the remote operator network can be realized. The source IP section range of the mobile terminal is not in the DDPU mobile terminal, and the network is accessed according to the original path planned by the operator and is not controlled by the DDPU.

The mobile terminal can access the public network or the local private network at the same time by writing the destination IP address field of the mobile terminal, and the destination IP address of the mobile terminal is written into the server address field distribution table through the network management M interface, so that the mobile terminal can simultaneously communicate with the local private network at the base station side and the public bearer network of the base station.

The three kinds of bidirectional communication are realized by modifying the distribution table through the network management interface without reconfiguring the network architecture and the operation and maintenance parameters, zero configuration of the data center, the server and the network at the base station side, zero configuration of the mobile carrier network of an operator, zero configuration of the base station and the router at the base station side and zero configuration of the mobile terminal are realized.

The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can make equivalent substitutions or changes to the technical solution according to the inventive concept of the present invention after seeing the technical solution disclosed in the present invention.