CN112217555A

CN112217555A - Formation satellite routing method based on SDN architecture and adopting SR routing protocol

Info

Publication number: CN112217555A
Application number: CN202010856002.6A
Authority: CN
Inventors: 李望; 陶孙杰; 祝佳
Original assignee: Chengdu Days Austrian Group Co ltd
Current assignee: Chengdu Days Austrian Group Co ltd
Priority date: 2020-08-24
Filing date: 2020-08-24
Publication date: 2021-01-12
Anticipated expiration: 2040-08-24
Also published as: CN112217555B

Abstract

The invention discloses a formation satellite routing method based on an SDN framework, belonging to the field of satellite communication; the invention designs a set of spatial network routing technical scheme based on an SDN framework and adopting an SR (sequence request) routing protocol, wherein the spatial network adopts the SDN framework, the main management function of the network is placed in an operation and management center on the ground, the interaction with users or services is completed, the network topology is saved, and the routing function is completed. The control function is placed in a ground gateway station, the control plane function is enhanced, and the idea of a distributed controller is combined. The satellite performs the switch function. On the user ground or space access side, the writing-in work of the source routing information is completed, and the subsequent nodes, whether on the satellite or on the ground, complete the forwarding by reading the source routing information in the data packet header without maintaining the forwarding information of each data stream. Therefore, most of the routing control of the space network can be finished on the ground without being restricted by on-satellite computing resources, so that the limited on-satellite resources are used for broadband data transmission, and the processing consumption is reduced.

Description

Formation satellite routing method based on SDN architecture and adopting SR routing protocol

Technical Field

The invention relates to the technical field of formation satellite network protocols, in particular to a routing method for meeting data service transmission of high and low orbit satellite nodes by formation satellite routing.

Background

The heaven and earth network routing is mainly used for solving the problem of data transmission between system sections and between users or nodes in the system sections, and based on a complex network topology structure, a reasonable optimal path is selected from multiple reachable paths between any source node and any destination node according to a certain rule, so that the indexes of network delay, throughput, link utilization rate and the like are ensured to meet the requirements.

Different from a ground network, a space segment network of the heaven and earth network consists of satellite nodes and inter-satellite links, and in order to ensure the reliability and performance of network operation, special design needs to be carried out on heaven and earth network routes according to the characteristics of the system by considering the characteristics of limited on-satellite processing and storage capacity, continuously and highly dynamically changed network topology, prolonged transmission time of the inter-satellite links, unbalanced data flow distribution and the like.

In the sixties of the last century, the first communication satellite has risen to space-the satellite communications business of mankind has taken an important first step. Through the development of half a century, satellite communication is continuously progressing: communication quality is continuously improved, and service range is wider and wider. With the advent of the 5G era, more communication modes are incorporated into the 5G communication framework, satellite communication is also regarded as an important component of 5G communication, and high-throughput satellites and new satellite constellations capable of meeting the 5G requirements can improve terrestrial services. As a supplement and a backup of ground communication, satellite communication can make up for the defect that a ground communication system is limited by the geographic environment, and can ensure instant communication in areas which cannot be covered by ground communication such as deserts, oceans and the like.

However, the satellite communication has the disadvantages of large time delay and high error rate because the satellite orbit is high and the transmission path with the ground terminal is long. In addition, the transmission path of the satellite network is complex, and the relative position between the satellite and the terminal and the atmospheric condition can influence the transmission. Inter-satellite handover is an unavoidable problem of low-orbit satellite networks: the constant movement of the satellite causes the user who is occupying the resource to leave the current satellite and access the next satellite. How to maintain communication during satellite handover and how to improve handover success rate and reduce handover delay are one of the key problems in spatial network research.

In 2008, Software Defined Networking (SDN) proposed by the professor of Nick attracted the line of sight of the industry and academia. In 2014, the idea of software-defined networking was introduced to satellite networks for the first time and software-defined networking structures in multi-layer satellite networks were proposed. The network structure is defined in a multi-layer satellite network. In this architecture, the low earth orbit satellite serves as the data plane entity and the geostationary satellite serves as the control plane entity. In addition, the software defined Network of the satellite separates the management plane, and a Network Operation and Control Center (NOCC) is used as a management plane entity. The structure is more suitable for the characteristics of a satellite network, the complexity of satellite nodes is reduced, and more complex functions are carried out on ground equipment.

The Software Defined Satellite Network (SDSN) has the greatest advantage of flexible control, and can realize coexistence of multiple protocols, thus becoming one of new solutions for Satellite networks in the 5G background. The flexibility of software defined satellite networks is a natural advantage over satellite routing and handover issues: the user or service can provide different QoS requirements, and the controller can select to use different routing and switching strategies according to the requirements. However, software-defined satellite networks should not be affected by the disadvantages of the mobility characteristics of satellite networks and the network model can be simplified by taking advantage of the periodic characteristics of satellite motion.

The software defined satellite network can enable the network to be flat and improve the flexibility of the network. Meanwhile, most satellite nodes only need to realize the function of table lookup forwarding, so that the complexity of the satellite nodes is greatly reduced. In addition, the flexibility of the flow table ensures that the forwarding function is not bound with hardware any more, thereby being convenient for the practice of a new protocol and reducing the development period of a new network function.

The end-to-end data transmission of the world network needs to span multiple heterogeneous network domains, and the establishment, maintenance and data transmission processes of a secure transmission channel can be seriously reduced by a channel environment with limited bandwidth and high delay. The relative positions of the heaven and earth network nodes are dynamically changed, and in order to ensure uninterrupted communication between the terminal nodes through a satellite or a neighboring space network, a safe switching mechanism is required to provide seamless network connection service. The space-ground network consists of a space-based network, a space-based network and a foundation-based network, wherein the space-based network consists of a satellite constellation network and has large communication link delay with ground network nodes. Since the heaven-earth network involves high orbit satellites, low orbit satellites and ground nodes, and relative motion exists among the nodes, the network topology is a dynamically changing process.

The software defined network is a centralized network, the controller contains network topology and resource utilization condition, a flow table can be flexibly generated according to the network condition and routing requirement, and the switch forwards the data packet according to the flow table. The software-defined satellite network usually comprises a management plane, a control plane is communicated with the management plane and a forwarding plane, a flow table of a forwarding basis is generated according to a strategy of the management plane and is sent to the forwarding plane, and meanwhile, information such as a link of the forwarding plane is monitored in real time. The forwarding plane has simple functions and mainly completes the function of forwarding the data packet according to the flow table.

The SDN controller is equivalent to the brain of a network, is responsible for uniformly controlling the forwarding equipment at the bottom layer, provides a regulation and control interface for an upper application service layer, and is located at a network key node in the SDN. Besides the southbound network control and the northbound service support, the controller also needs to pay attention to the east-west extension and the extension of a distributed controller cluster, so that the bottlenecks of safety and performance caused by the centralized control of the SDN can be avoided.

The core concept of the SDN is to separate a control plane of a network from a data plane of a user, transfer an original network control function into a centralized controller, replace the control plane in a switching device by a standardized southbound interface, and provide a programmable northbound application interface for the controller to be called by an upper layer user.

The network control technology of the controller mainly comprises the steps of link topology discovery, topology framework management, management strategy formulation, flow table item issuing and the like through a southbound interface protocol. The link topology discovery and topology architecture management mainly comprises that a controller obtains connection information of bottom layer switching equipment through an uplink channel of a southward interface to realize the functions of link discovery and topology management of the controller, and the controller implements a technology of uniformly controlling network equipment through a mechanism of formulating a strategy and issuing flow table items through a channel of a connection switch of the southward interface.

The SDN controller realizes the functions of a link discovery protocol by utilizing an LLDP protocol. The LLDP protocol is a link discovery protocol, and can be distributed to its own directly connected network nodes by forming different TLVs, i.e. types, lengths and values of data packets, encapsulated in LLDPDU, in link layer discovery protocol data units, and these network nodes store these information in the form of standard management information base, providing a database for querying and judging link communication status for the network management system.

The topology management function of the SDN is a core problem to be solved by a controller of a spatial network. Topology management mainly has two functions, namely monitoring and collecting network link connection conditions and recording logic networking information in real time. The former is to monitor and collect the working condition of the SDN switch in the network in real time, collect the equipment working and link connection state information of the network in time, and store the information in the database. For the time-varying topology network, the controller needs to regularly send LLDP protocol data packet information to the adjacent network nodes, and then collect the acknowledgement packet fed back, so as to monitor the working state of the switch and update the network topology architecture view in time. The second function is to perform network service with different functions according to the requirements of different users. And in the logic networking, on the basis of acquiring the topology of the existing network, optimizing the network resources again according to the network resources required by the user, and allocating the network resources. The SDN controller can also support the virtualization technology of the tenant network, isolation is carried out on the aspects of performance, safety and the like, and better tenant network experience is provided.

Another important function of the SDN controller is policy making and flow entry issuing. And the SDN controller controls the SDN switch by issuing the flow table entry. The SDN switch selects a forwarding path according to a flow table item issued by the controller, which is equivalent to a routing table of the existing network.

The most important characteristic of the flow table mechanism of the SDN is that the traditional network hierarchical management is broken. The SDN controller realizes the uniform identification and forwarding of the multi-level network by uniformly encapsulating the packet header of the MAC protocol, the packet header of the IP protocol and the packet header field of the TCP/UDP protocol into a flow table item. Therefore, protocols of different protocol layers in the traditional network can be identified through the routing table and corresponding actions are executed. Therefore, the controller needs to formulate a corresponding forwarding policy according to network transmission requirements of different layers, generate a corresponding flow entry, and send the flow entry to the SDN switch.

The SDN controller needs to implement the following functions:

and the working state and the communication event of each network node in the network are automatically discovered. The communication events comprise the change of the overall topological structure of the network, the addition, the exit and the disconnection of equipment in the network, the connection change of links of each node and the balanced load of flow in the network.

The communication between network device nodes can be controlled, including between controllers and switches, for example, Openflow protocol, which can support the communication protocol between network nodes.

Devices and shared resources of a network are managed. Including the use of memory space, load scheduling of threads of communication events in the controller, and testing the performance of various network nodes and communication links, among others.

And providing visual tools and network configuration services for the application layer. A general visual tool refers to reflecting the overall network topology through a web interface. The network configuration service means that a user can perform specific network resource allocation according to actual network requirements, for example, a data packet including a certain field can be instructed to be forwarded to a certain switch and other specific functions.

The bottlenecks and risks faced by the formation satellite network SR are as follows:

in a formation satellite network, the link quality difference of the ground, low-orbit constellation and high-orbit constellation is large, especially the link between the satellites is frequently changed, the quality fluctuation is large, and the fluctuation period is different, if a classic SR technology is adopted, the PMS is required to obtain accurate and real-time network state information, and under the environment of the formation satellite network, the transmission delay of hundreds of milliseconds is enough to cause the second-level lag between the network state obtained by the PMS and the actual network state.

The research and development of the satellite node data plane are limited by various special processes and environments, the supported label stacking quantity may not have good expandability like a ground router, and the dilemma that the label stack cannot contain all routing information easily occurs by completely depending on a source routing technology.

Since the spatial network topology changes frequently, one or more paths including multiple hops at the IP layer cannot be directly multiplexed by using the SR technique, as in the terrestrial network, but pushing each node label into the data packet header would have the first and second problems, and would cause a large packet header overhead in the spatial network with valuable bandwidth.

Disclosure of Invention

The invention discloses a satellite routing method based on an SDN framework and formed by adopting an SR (sequence request) routing protocol, and aims to provide a heaven-earth network routing method which is suitable for the characteristics of limited on-satellite processing and storage capacity, continuous high-dynamic-change network topology, long transmission time of inter-satellite links, unbalanced data flow distribution and the like.

The technical scheme of the invention is a formation satellite routing method based on SDN architecture and adopting SR routing protocol, and the system aimed by the method comprises the following steps: user terminal, satellite, gateway station, operation control center; the user terminal is communicated with the satellite, the satellite is communicated with the gateway station, and the gateway station is communicated with the operation and control center; grouping the satellites according to different orbital planes, configuring a gateway station with a controller in each group, and exchanging information between the gateway station of the controller and all the satellites on the orbital planes to finish generating a flow table; the controller gateway stations communicate through a ground link, the obtained network information is forwarded to other controller gateway stations, and each controller gateway station has full network information; the satellite, the gateway station and the user terminal support an OpenFlow protocol and are provided with southbound interfaces capable of communicating with the controller; supporting an Of-Config protocol for configuring and managing the network device; the method has the advantages that safe transmission between the controller gateway station and the adjacent transmission nodes is guaranteed, the flow table can be received from the controller gateway station, and network information including link states and resource utilization conditions is collected by forwarding data packets according to the flow table;

the routing method comprises the following steps:

step 1: the controller gateway station sends the satellite-based node motion track information to an operation and maintenance control center and responds to a routing request;

step 2: the operation and maintenance management and control center constructs a virtual topological graph;

and step 3: the operation and maintenance management and control center establishes a source routing table;

and 4, step 4: the operation and maintenance control center injects the routing table to the satellite node through the controller gateway station;

and 5: the on-board nodes accept the tasks to construct task routes, and the routes are carried out according to the source routes carried by the data packets; if the packet transmission is successful, entering the next step, if the packet transmission is unsuccessful, maintaining a routing table, and reconstructing a task route;

step 6: completing service packet transmission;

and 7: waiting for the next task;

the gateway station of the controller sets service flow according to the configuration and transmission capability of different network nodes including the satellite and aiming at the QoS requirements of different transmission services, configures a flow table entry, constructs a transmission path meeting the QoS requirements of different transmission services, and completes routing and link maintenance; different network nodes including the satellite synchronize flow table configuration with the controller and upload network state information to realize special data transmission;

the controller in fortune dimension management and control center includes: the system comprises a storage module, a global control layer, a cluster controller selection module, an inter-neighbor controller discovery module, a device management module, a link discovery module, a topology management module, a topology service module, an SDN switch, an Openflow switching module, an application layer interface and an application layer, wherein the storage module stores an all-node topology structure and an application management strategy, the storage module performs data exchange with the device management module and the application layer interface, the device management module performs data exchange with the inter-neighbor controller discovery module link discovery module respectively, the inter-neighbor controller discovery module performs data exchange with the cluster controller selection module and the topology service module, the cluster controller selection module performs data exchange with the global control layer, and the topology service module performs data exchange with the SDN switch; the link discovery module is connected with an application layer interface through an Openflow exchange module, and the application layer interface is connected with the application layer; the Openflow switching module comprises: module management, thread pool, Openflow protocol, Web interface, the said application layer interface includes: forwarding policy and today matching domain entry;

the route discovery module detects the link by adopting a link discovery protocol and a broadcasting mode; sending a Packet-out data Packet by using a link discovery protocol to obtain a Packet-in Packet sent by adjacent equipment, so that the information of the directly connected equipment can be known; sending a Packet-in Packet to the adjacent equipment, and requiring the adjacent equipment to broadcast the Packet, wherein the method can pass through a switch which does not belong to the controller to reach the next Openflow switch, and then the switch can send a Packet-in data Packet to the controller, so that the controller can create an integral network topology view;

after the link is found, the equipment management module sends a request packet, learns the characteristics of the connected equipment, classifies the equipment according to the entity classifier, and defines equipment by adopting a mac address and a vlan address; in the formation satellite network, the zone bit is read, the equipment type is judged, whether the equipment is a switch or a controller is judged, and then the next step of judgment is carried out;

the cluster controller selection module completes the protocol stack task of the controller cluster and completes the information sharing among the controllers in the layer.

Further, a fast rerouting method based on the routing method includes:

step 1: distributing fixed labels for all network elements in the network by using MPLS labels; the tag management table consists of: the special virtual line comprises a special virtual line ID, an IP quintuple, a special virtual line mode, a special virtual line category and a corresponding special virtual line label;

step 2: establishing a virtual private line, and dividing the virtual private line into a user-level virtual private line and a backbone-level virtual private line; the user-level virtual private line is deployed in an end-to-end mode and directly serves the user; the backbone level virtual private line is deployed in a special area in the network and is used for ensuring the transmission quality and reliability between two forwarding nodes in the network; in the data transmission process, the data packet header only contains a label corresponding to the requirement, and a forwarding rule aiming at the label is issued at nodes along the way;

establishing three virtual special line routing methods on the basis of the step 1 and the step 2: a single label virtual special line routing method, a nested mode virtual special line routing method and a splicing mode virtual special line routing method;

the single label virtual special line routing method comprises the following steps: PMA at the entrance of the virtual private line is a unique identification of the whole network corresponding to the service flow encapsulation purpose IP, the function of the layer of labels is that after the virtual private line has a problem and the corresponding labels are stripped, the data packet can still be forwarded according to the default route, and the minimum network connectivity is ensured by using an IP routing protocol; then PMA packages second layer MPLS label, which is virtual special line label to make business flow be transferred according to virtual special line to realize flexible dispatching; in the subsequent forwarding routing process, the whole virtual private line uses the same MPLS label as an identifier to forward;

the nested mode virtual special line routing method is characterized in that another layer of virtual special line is nested in one layer of virtual special line, and the specific method comprises the following steps: PMA at an entrance in a user virtual private line is a unique identification of the whole network corresponding to a service flow encapsulation purpose IP, the function of the layer of labels is that after the virtual private line has problems and the corresponding labels are stripped, data packets can still be forwarded according to a default route, and an IP routing protocol is utilized to guarantee the minimum network connectivity; then PMA encapsulates the second layer MPLS label, the label is the label of virtual special line, to make the service flow forward according to the virtual special line and enter the backbone virtual special line, the PMA at the entrance of the backbone virtual special line encapsulates the third layer MPLS label for the service flow, the layer label is the label scheduled in the backbone virtual special line;

the splicing mode virtual special line routing method is characterized in that a plurality of paths are spliced into one path; the specific method comprises the following steps: the backbone virtual private line is regarded as a section of virtual link, and the user end-to-end virtual private line is regarded as three sections of splicing of a source site access link, the backbone virtual private line and a target site access link; three corresponding marks are directly pressed into the header of the user service flow in sequence, the outermost layer is a mark corresponding to the source station access link, the second layer is a mark corresponding to the backbone virtual private line, and the bottom is a mark corresponding to the target user station access link.

The invention has the beneficial effects that:

1. the invention designs a set of spatial network routing technical scheme based on an SDN framework and adopting an SR (sequence request) routing protocol, wherein the spatial network adopts the SDN framework, the main management function of the network is placed in an operation and management center on the ground, the interaction with users or services is completed, the network topology is saved, and the routing function is completed. The control function is placed in a ground gateway station, the control plane function is enhanced, and the idea of a distributed controller is combined. The satellite performs the switch function.

2, the operation and maintenance control center with stronger calculation processing capacity is adopted to construct source Routing information, an original Routing table is created, the same Routing table is merged and simplified, the ground operation and maintenance control center injects the created Routing table to each satellite-based node, the satellite-based node uses a Segment Routing mark to route when service grouping arrives, and dynamic Routing table maintenance is carried out when necessary, so that the calculation overhead and the processing overhead of the satellite-based node can be greatly reduced. The problem of high cost for constructing the routing topology on the satellite due to limited power, limited computing resources and weak processing capability of the satellite-based nodes is solved by the upper note routing table.

3. In the Routing stage, the on-satellite load task is obtained, and a data packet is sent by adopting a source route constructed by Segment Routing; the data packet transmission is completed, the data service transmission of high and low orbit satellite nodes is met, the next task is waited, the resource consumption caused by the satellite routing calculation is reduced, and the problem that the satellite real-time routing capability is limited due to the limited computing resource and processing capability of the satellite-based nodes of the world network is solved.

Drawings

Fig. 1 is a schematic diagram of a Routing flow of a satellite Routing method based on Segment Routing formation according to the present invention.

Fig. 2 is a formation satellite network system based on an SDN architecture.

Figure 3 is a formation satellite network architecture based on an SDN architecture.

Figure 4 is a generic controller architecture based on SDN.

Fig. 5 is an Openflow flow entry.

Figure 6 is an Openflow flow table workflow.

Figure 7 is a SDN formation based satellite network controller design.

Fig. 8 is a schematic diagram of MPLS label pool partitioning.

Fig. 9 is a schematic diagram of a single tag virtual private line principle.

Fig. 10 is a schematic diagram of the working principle of the nested mode virtual private line.

Fig. 11 is a schematic diagram of a virtual private line principle in a splicing mode.

Fig. 12 is a schematic diagram illustrating a principle of fast virtual private line switching.

Fig. 13 is a schematic diagram illustrating a principle of fast switching of a virtual private line to an alternate line.

Detailed Description

In order to achieve the purpose, the invention designs a set of technical scheme of spatial network routing based on an SDN architecture and adopting an SR (sequence request) routing protocol, wherein the spatial network adopts the SDN architecture, the main management function of the network is an above-ground operation and management center, and meanwhile, the interaction with a user or a service is completed, the network topology is stored and the routing function is completed. The control function is placed in a ground gateway station, the control plane function is enhanced, and the idea of a distributed controller is combined. The satellite performs the switch function. On the user ground or space access side, the writing-in work of the source routing information is completed, and the subsequent nodes, whether on the satellite or on the ground, complete the forwarding by reading the source routing information in the data packet header without maintaining the forwarding information of each data stream. Therefore, most of the routing control of the space network can be finished on the ground without being restricted by on-satellite computing resources, so that the limited on-satellite resources are used for broadband data transmission, and the processing consumption is reduced.

The SR routing generation flow is shown in fig. 1, and the flow is as follows:

step 1: the gateway station sends the satellite-based node motion trail information to an operation and maintenance control center and responds to a routing request;

step 2: entering an SR route generation stage, firstly, constructing a virtual topological graph by an operation and maintenance management and control center;

and 4, step 4: the operation and maintenance control center injects the routing table to the satellite node;

and 5: entering a dynamic routing stage, the satellite nodes accept tasks to construct task routes, and routing is carried out according to source routes carried by data packets; if the packet transmission is successful, entering the next step, if the packet transmission is unsuccessful, maintaining a routing table, and reconstructing a task route;

step 6: completing service packet transmission;

and 7: waiting for the next task.

A three-layer control structure of a management plane, a control plane and a data plane is adopted, wherein the management plane is an application defined by a user or a service and is communicated with the control plane only when a network protocol or user requirements change. The control plane stores the topology state of the whole network and the strategy of the management plane, and the routing and other functions are completed according to the strategies. At the same time, the control plane remains in communication with the data plane through the southbound interface.

The scheme adopts a control strategy that low orbit satellites are grouped, and each group is controlled by one controller. The strategy places controllers at different gateway stations, adopts the idea of a distributed controller, groups satellites according to different orbital planes, and adopts one gateway station in each group as a controller to exchange information with all satellites on the orbital planes so as to complete the functions of generating flow tables and the like. The strategy has the advantage of shortening the propagation path from the satellite switch node to the controller. In general, when a user accesses a satellite or sends a first packet, the satellite switch does not have a corresponding flow table and needs to send a request to the controller. By shortening the propagation path, the waiting time of the user when the first data packet is sent can be reduced, and the propagation delay is reduced.

The control structure focuses on the data plane and the control plane: the satellite works on a data plane as a switch to complete the functions of forwarding data packets, collecting link information and the like; the gateway station has the function of a controller, works in a control plane, and needs to complete the functions of calculating a route and the like. Each orbital plane is provided with a gateway station with a controller function, which manages all the satellites in the same orbital plane. The controllers communicate through a ground link, and the obtained network information is forwarded to other controllers, and each controller has full-network information.

According to the configuration and transmission capability of different network nodes including the satellite, the controller sets service flow according to the QoS requirements of different transmission services, configures flow table items, constructs a transmission path meeting the QoS requirements of the different transmission services, and completes routing and link maintenance; different network nodes including the satellite synchronize flow table configuration with the controller and upload network state information, special data transmission is achieved, the load on the satellite in the satellite network is reduced, and the data transmission efficiency is improved. The network system comprises two parts of a control plane and a forwarding plane, and the network architecture is shown in fig. 2.

The SDN controller is deployed in an operation and maintenance management and control center and is responsible for managing and controlling the whole network, and each satellite of the low earth orbit satellite layer is an OpenFlow switch which only has a forwarding function in the network. Routes and links can be reasonably configured on the basis of the routing.

The control strategy comprises three parts, namely a management plane, a control plane and a data plane. The management plane is the user-defined interface and the contents of service, etc., and is connected with the controller through the northbound interface. The north interface of the management plane is mainly a programmable logic interface, is convenient for obtaining new service requirements and user-defined services, is arranged in the operation and maintenance management and control center and is convenient for communicating with ground users. The plane is consistent with the ground software defined network and does not need to be functionally improved. The management plane communicates with the satellite only when service requirements, etc., change, and normally does not affect the network. The environment in the scenario assumes that the network has stabilized and that the various protocols and network requirements have not changed. The control structure focuses on the data plane and the control plane: the satellite works on a data plane as a switch to complete the functions of forwarding data packets, collecting link information and the like; the operation and maintenance control center has the function of a controller, works on a control plane, and needs to complete the functions of calculating routes and the like.

As shown in fig. 3, the satellite network includes a satellite ground gateway station, an operation and maintenance control center, and a satellite node. All the satellite nodes and the gateway stations serve as switches to work on a data plane, and the operation and maintenance control center comprises application and control functions and is arranged on a control plane and a management plane. The management plane is a functional level, and is usually integrated into the controller to serve as a bridge for users or services to communicate with the controller. And when a user has a new service requirement, writing a new application through the programmable interactive interface. The functions of the layer are integrated in an operation and maintenance management and control center, and once a user has new requirements or adopts a new protocol, the strategy is updated through the satellite-ground link connection controller. In general, functions such as routing and forwarding of a satellite network do not involve a management plane.

The different planes function as shown in figure 4.

The data plane entity comprises a satellite node, a ground satellite gateway station and a user terminal. The data plane functions include:

1) the system supports an OpenFlow protocol and is provided with a southbound interface capable of communicating with the controller;

2) supporting an Of-Config protocol for configuring and managing the network device;

3) the method has the advantages that safe transmission between the controller and the switch is guaranteed, the flow table can be received from the controller, and the data packet is forwarded according to the flow table, and the flow table is the core for the formation satellite software to define the network to forward the data flow;

4) network information is collected including link status, resource utilization, etc. The information is reported to the control plane device for generating a policy according to the network condition.

The OpenFlow protocol includes three information types, which ensure smooth communication between the controller and the switch, and the Of-Config protocol specifies how to configure the network device. Thus, supporting both protocols ensures the ability of the controller to manage the data plane.

The control plane entities are ground gateway stations with controller functions, and these nodes perform two main functions of the control plane, namely, managing the whole network state and calculating the routing strategy. The controller node obtains the state information of the whole network from the report sent by the switch satellite; and obtaining the current network topology and the link use condition from the information, calculating a route according to the information and the user requirement, making the path information into a flow table, and managing the inter-satellite link resource information such as beams. And selecting an operation and maintenance control center in the network as a super controller, wherein the super controller contains the whole network information and manages the whole network resources.

A SDN based pass control architecture is shown in fig. 5.

For the system design of the controller, the following functional tasks need to be accomplished:

designing a north communication interface from a controller to an application layer, and designing a control mode of a user to the controller, wherein the control mode comprises the design of the requirement of the user on an SDN network, the cloud computing technology of realizing virtualization of the SDN and the like.

The controller system is connected to the southward communication interface of the forwarding device, the communication interface realizes the functions of link discovery, device maintenance, flow table item issue and the like of the forwarding device by the controller, and simultaneously realizes the feedback of data packets, control information application and the like from the forwarding device to the controller.

In the controller system, an east-west direction interface for communication between the controller and the controller needs to be designed, and a good east-west direction communication interface design is beneficial to improving the control capability of the controller and improving the expansibility of the controller.

The controller is a core component of the SDN network, and the controller system is composed of a large number of application examples.

The southbound interface protocol is an Openflow protocol; the Openflow protocol realizes core tasks such as link discovery, equipment management, flow table item making and issuing and the like. In the Openflow protocol, a forwarding rule that specifies a network device is specified by a flow table, and the forwarding device processes a packet in accordance with the flow table. And a special Openflow channel is adopted between forwarding equipment and a controller in the network to transmit information. The flow table entry may be specified by an upper layer application, or may be automatically made and issued by the controller according to the link condition of the network.

The header field of the flow table entry can be fused with the network protocols of all layers in the seven-layer network protocol, and a new flow table entry is designed aiming at the protocol current situation in the formation satellite network.

A flow table may be regarded as an abstraction of Openflow for the data forwarding function of the network device. In a conventional network device, data forwarding of a switch depends on a two-layer MAC address forwarding table stored in the device, a router needs a three-layer IP address routing table, and the same is true of a flow table used in an Openflow switch. Different from the above, the conventional network device can only support one layer of network configuration information, but network configuration information of each layer in the network is integrated in the Openflow flow entry, so that richer rules can be used when data forwarding is performed, and fig. 6 is a work flow diagram of the Openflow flow entry.

The SDN network architecture supports zero to multiple action processing of data packets and supports processing of the same data packet by multiple devices. And there may be a prioritization of the execution of the matching of different devices and different flow entries. But the transmission of the data packets cannot guarantee their precedence. In addition, for some data packets, all the flow table entries cannot match them, and the switch will pass the data packet through the channel to the controller, and the controller determines the processing of the data packet.

Fig. 6 shows an execution flow that the Openflow switch needs to expand after receiving a network packet of a communication node. The Openflow protocol supports a centralized control mode, and the SDN controller uses the Openflow protocol, creates a topology view by collecting network devices and link connection states, and then issues a flow table to the switch, so that a channel for transmission between the controller and the switch is very important. The controller manages and controls the Openflow switch through the transmission channel interfaces, and receives link information or data packets transmitted by the Openflow switch through the interfaces.

The east-west interface refers to a communication interface between the controller and the controller.

In the scheme, the east-west interface of the controller is divided into an intra-layer east-west interface and an inter-layer east-west interface.

The east-west interface between layers refers to controller communication between controller clusters. For this part of the communication, the openflow protocol is followed. The other controllers are considered as one device.

The east-west interface in the layer refers to communication among controllers in the controller cluster, is used for realizing control information sharing of the same control cluster, and is beneficial to managing communication equipment connected with the same controller. In the controller cluster, Jgroups are used to realize information sharing among controllers.

And aiming at the set hierarchical cluster controller system, designing a functional module of the formation satellite network controller.

The controller generally needs to have the following functions:

1) the communication transmission between the controller and the switch is realized, and modules such as flow table issuing, flow control and the like are completed;

2) thread management, namely completing module shared resource management;

3) and a controller interface is provided for a user, a Web UI is provided, and friendly debugging service is improved for the user.

For the formation satellite network controller, besides the above functions, a link discovery, thread communication, and resource sharing mechanism between the controller and the controller need to be implemented. The improved controller function module is shown in fig. 7.

SDN controllers need to implement control functions such as link discovery, device communication control, process management, and user debugging interface provision. In the formation satellite network, in addition to the device identification and device communication between the controller and the switch, the device identification, device communication and resource sharing between the controller and the controller need to be realized.

For network controller design, the same protocol from controller to switch is used for communication protocol between controller and controller. Taking the Floodlight controller as an original model, and in order to adapt to the satellite network environment, carrying out the following changes on the original functional modules in the Floodlight:

1) link discovery module

The link discovery module probes the link using LLDP, a link discovery protocol and broadcast. Sending a Packet-out data Packet by using an LLDP protocol so as to obtain a Packet-in Packet sent by adjacent equipment and know the information of the directly connected equipment; the method can pass through a switch which does not belong to the controller to reach the next Openflow switch, and then the switch can send a Packet-in data Packet to the controller, so that the controller can create an overall network topology view.

For spatial network control, the link discovery module discovers not only the topological locations of the adjoining switches, but also the topological locations of the neighboring controllers. Therefore, when designing a Packed-in receipt packet, a flag bit of a controller needs to be added and the controller devices need to be processed separately.

2) Device management

After the link is found, the device manager sends a request packet, learns the characteristics of the connected devices, and classifies the devices according to the entity classifier. A device is typically defined using mac addresses and vlan addresses. However, in the formation satellite network, in order to enable better network device management, the flag bit is read, the device type is judged whether the device is a switch or a controller, and then the next judgment is performed.

3) Memory module

By using an NIB (network information base) data model of the Onix controller for reference, information sharing is realized in the controller cluster. Each controller is provided with a storage module for storing the information of the communication nodes independently connected with the controller, and the Jgroups protocol of the cluster is utilized to realize the resource sharing of each controller device in the cluster, so that the topology information of the whole network is acquired, and more reliable information is provided for the forwarding of the communication nodes.

4) Inter-neighbor controller discovery

Because the relative position of the interlayer controller is relatively fixed, the topology does not change greatly, but the relative position between different layers can change periodically. The inter-neighbor controller discovery is mainly for controllers between different layers. Each controller needs to continuously maintain the connection state to the controllers of different layers even if the connection state can be maintained with one controller of different layers due to the rapid change of the relative positions of different layers.

5) Cluster control

The module completes the protocol stack task of the controller cluster and completes the information sharing among the controllers in the layer, and the module is mainly designed according to Jgroups.

Fast reroute based on SR

A hybrid SR scheme will be employed. In the scheme, on one hand, fixed resource identifiers are distributed to all network elements in the network by using MPLS labels, and a classic SR technology is supported, so that the PMS can calculate and call when issuing virtual private lines. On the other hand, the virtual private line mode of a single label identification is supported, the data packet header does not encapsulate the whole source routing information any more, but only contains a label corresponding to the requirement, and a forwarding rule aiming at the label is issued at the nodes along the way so as to be compatible with the heterogeneous data plane.

On the basis, the virtual private line is divided into a user-level virtual private line and a backbone-level virtual private line, the user-level virtual private line is deployed in an end-to-end mode and directly serves users, and the backbone-level virtual private line is deployed in a special area in a network and used for guaranteeing the transmission quality and reliability between two forwarding nodes in the network and providing public service quality guarantee for all user traffic passing through the area. Especially in some network areas with poor stability and larger user flow, the preset backbone level virtual special line can effectively shield instability of the underlying network to the upper layer service and the PMS, and the instability is abstracted to a virtual link, and the maintenance of the backbone level virtual special lines can adopt a denser mode and more redundant resource supply, so that the maintenance cost is reduced from N of the user level virtual special line to 1 of the backbone level virtual special line. Because the maintenance of the backbone virtual private line is also locally executed at the starting point and the end point, the maintenance efficiency is far higher than that of the inlet node and the outlet node of the remote user virtual private line.

The SR scheme proposed by the present invention involves the following:

the whole data plane is divided into a backbone network and a user access network, and the PMA is used for identifying the network type to the controller PMS, so that the aggregation of the repeated part of the virtual private line is realized. The repeated parts in the virtual private lines are aggregated to be used as a public virtual private line in a backbone network, and the virtual private line of a user is encapsulated, so that the maintenance amount and the number of flow table entries on a forwarding node can be reduced, the calculation and issuing deployment pressure of a controller PMS and the flow table query cost of the forwarding node are reduced, and the forwarding performance of a control plane and a data plane is indirectly improved;

on the basis of carrying out type division on the nodes, a two-layer nesting mode of a backbone virtual private line and a user virtual private line is adopted, and the effect similar to LSP aggregation in an MPLS network is obtained. The virtual private line in the backbone network is maintained for a long time, and the forwarding capability and the maintenance function of the backbone network part are provided for the virtual private line of a user. The calculation of the user-level virtual private line regards the backbone virtual private line as a virtual link, does not concern which sites the user-level virtual private line specifically passes through, and only concerns the starting point and the end point of the user-level virtual private line, so that the calculation center of gravity of the user-level virtual private line becomes the problem of calculating which backbone forwarding site is connected from a user access site and selecting which backbone virtual private line;

three virtual private line modes are designed: the system comprises a single label virtual private line, a nesting mode virtual private line and a splicing mode virtual private line. And flexibly scheduling the user service flow aiming at different network scenes so as to ensure the quality of QoS.

The virtual private line maintenance function is divided into the following points:

and establishing a virtual private line. In the formation satellite network, through the data collected by the management plane and the calculation of the control plane strategy, personalized virtual private lines can be issued according to different requirements, so that data streams can be forwarded from different links according to the network condition. Not only QoS is guaranteed, but also the network utilization rate is improved. The PMA shall be able to issue virtual private lines of various modes to a spatial or ground rice grabbing node according to a path calculated by the centralized controller, so as to ensure the establishment of the virtual private lines;

and (4) canceling the virtual private line. And if the issued virtual private line does not meet the requirement, the issued virtual private line needs to be cancelled, and a new virtual private line meeting the requirement is issued. The revocation function of the virtual private line is also necessary. The distributed agent can cancel the virtual private line which does not meet the requirement or is redundant in real time according to the instruction of the centralized controller;

and encapsulating the service flow header according to the virtual private line. The virtual private line needs to encapsulate the MPLS label for the service flow, so that the service flow can be matched according to the established virtual private line and forwarded. Nodes in the network carry out the encapsulation and decapsulation operations of the MPLS labels on the headers of the service flows through the flow tables, so that the service flows have various identifications, and then enter the virtual private line to be scheduled.

Details of the virtual private line tag management technique are shown in table 1. The label management table of the virtual private line consists of a virtual private line ID, an IP quintuple, a virtual private line mode, a virtual private line category and a corresponding virtual private line label list. The mode field in the table represents the mode of operation of the virtual private line and the IP quintuple represents the object of the service. The label list comprises two fields of a label value and a label type, wherein the label type is all MPLS at present, but other label identifications can be used in the future from the viewpoint of expansibility, and therefore, the label type field is designed.

The MPLS label pool partitioning rule is shown in fig. 8 below.

The label pool has 5 sections which respectively correspond to the following sections:

1) an MPLS label pool of IP mapping of nodes and links in a management plane and a control plane in a formation satellite network;

2) an MPLS label pool corresponding to the requirement ID of the user virtual private line;

3) an MPLS label pool corresponding to the requirement ID of the backbone virtual private line;

4) MPLS label pool corresponding to the user virtual private line;

5) and MPLS label pools corresponding to the backbone virtual private lines.

The virtual private line requirement ID is an ID unique to the entire network for identifying a user requirement when QoS is guaranteed for a specific user traffic.

The MPLS label range corresponding to the requirement ID of the backbone virtual private line is 100,001-. And the MPLS label value corresponding to the demand ID of the backbone virtual private line is the same as the demand ID.

The MPLS label range of the requirement ID of the user virtual private line is 102,001-. And the MPLS label value corresponding to the requirement ID of the user virtual private line is also the same as the requirement ID of the user virtual private line.

Each virtual private line corresponding to the demand ID corresponds to a dynamic MPLS label. For each demand ID, 20 MPLS labels are allocated as the virtual private line MPLS label corresponding to the demand, and since one path is bidirectional from one path to one path, it means that 10 virtual private lines can be supported at most. Starting at 200,000, 20 are grouped together. The number of demand IDs is 1 ten thousand. The corresponding virtual private line MPLS label is therefore 20 thousands, with each 20 consecutive numbers in a group serving only his demand ID.

The MPLS label range of the backbone virtual private line is 200,001-240,000. There are 4 thousands, starting with 200,001, of which every 20 consecutive numbers correspond to a backbone level requirement. Then each demand corresponds to MPLS labels ranging from (ID-100,001) × 20+200,001 to (ID-100,001) × 20+200,000+ 20.

The working principle of the single label virtual special line is as follows:

in the single label virtual private line mode, the whole virtual private line uses the same MPLS label as the identifier to forward. As shown in fig. 9, in the communication between two terminals, the original header of the data packet is 1.1.1.1 of the source IP, and the destination IP is 2.2.2.2.

The PMA at the entrance of the virtual private line firstly packages the unique identifier '1' of the whole network corresponding to the IP of the destination for the service flow, and the label of the layer has the function that the data packet can still be forwarded according to the default route after the virtual private line has a problem and the corresponding label is stripped, so that the minimum network connectivity is ensured by utilizing an IP routing protocol. This layer label is also referred to as a life buoy layer label, and is hereinafter referred to collectively as a life buoy layer label.

Then, the PMA encapsulates a second-layer MPLS label "3", which is an identifier of the virtual private line, so that the service flow can be forwarded according to the virtual private line, thereby implementing flexible scheduling. In the single label mode, the whole virtual private line is forwarded by taking a single label as an identifier.

The PMA at the virtual private line exit will decapsulate the two-layer MPLS label and recover the original data header.

The working principle of the nested mode virtual special line is as follows:

the nested mode is that another layer of virtual special line is nested in one layer of virtual special line, namely, a layer of backbone virtual special line is nested in the user virtual special line. The user networks at the two ends are guaranteed by user virtual private lines, and the middle backbone network is guaranteed by a backbone virtual private line. In the nested mode, the user virtual private lines at the two ends are actually the same virtual private line, and the same virtual private line identifier is used for maintaining the same virtual private line.

In this mode, the user virtual private line is still identified with a single label, the same as the single label mode, but the portion that traverses the backbone network is nested into the backbone virtual private line. Thus, the PMA at the entrance of the backbone virtual private line encapsulates a third layer MPLS label for the traffic flow, which is the identifier scheduled in the backbone virtual private line. In this scenario, the routing table generated by the IP routing protocol still serves to provide minimal security.

The operation of the user data packet in nested mode is shown in figure 10. The original data packet header is source IP 1.1.1.1, and the destination IP 2.2.2.2. A backbone virtual private line with an MPLS label of "4" is pre-established between the forwarding nodes B, C.

The PMA at the entrance of the user virtual private line firstly packages a life ring layer label '1' for the service flow. The PMA will then encapsulate a second layer MPLS label "3", which is a customer virtual private line label, so that the traffic flow can be forwarded in the customer virtual private line.

When the service flow reaches the backbone virtual private line entry node B, the PMA at the position packages a layer of backbone virtual private line MPLS label '4' for the service flow according to the stack top label '3', and the layer of label is the label of the backbone virtual private line, so that the service flow is forwarded by a section of path in the backbone network.

After the virtual private line reaches the backbone virtual private line exit node C, the PMA at that location will remove the MPLS label "4" of the backbone virtual private line, so that the service flow is continuously forwarded in the second half of the user virtual private line.

After the user virtual private line exit is reached, the PMA at the position can decapsulate the remaining two layers of MPLS labels, and recover the original data header.

The working principle of the splicing mode virtual special line is as follows:

the splicing mode is to splice several paths into one path as the name implies. The backbone virtual private line is regarded as a section of virtual link, and the user end-to-end virtual private line is regarded as three sections of splicing of a source site access link, a backbone virtual private line and a destination site access link. Referring to the source Routing mode in Segment Routing, three corresponding marks are directly pressed into the header of the user service stream in sequence, the outermost layer is a mark corresponding to a source site access link, the second layer is a mark corresponding to a backbone virtual private line, and the bottom is a mark corresponding to a target user site access link.

In this mode, the backbone virtual private line is visible to the source site PMA, so the entries in the tag label management table issued by the PMS to the source site PMA are no longer single tags, but are tag stack information in groups of three. When the splicing mode is adopted, if the OVS is adopted as data plane equipment, life ring layer marks cannot be encapsulated for service flow, if a matched mark forwarding table cannot be found in the intermediate node, the data packets can only be discarded, and the life ring marks on the bottom layer cannot be forwarded according to an IP routing table as in a nested mode.

The operation of the user data packet in the splicing mode is shown in fig. 11, and the two terminals communicate with each other, and the original data packet header is a source IP of 1.1.1.1, and the destination IP is 2.2.2.2. The virtual private line passes through four nodes A, B, C, F, the label of the user virtual private line is "3", the backbone virtual private line is preset between the BC, and the requirement ID is "212314".

The PMA at the entry of the virtual private line will encapsulate the MPLS label "3" of the virtual private line at the opposite end for the service flow. The PMA will then encapsulate the second layer MPLS label "212314", and the third layer MPLS label "3" of the local virtual private line. The second layer label "212314" is a required ID number of the backbone virtual private line, and is not an MPLS label used by the actual backbone virtual private line.

Because the head node and the backbone virtual private line entry node are not the same point, the PMA of the head node cannot know the MPLS label number currently used by the backbone virtual private line at the moment, and only the label of the required ID of the backbone virtual private line is encapsulated first, and then the actually used MPLS label is mapped back at the backbone virtual private line entry. Mapped to label "4" by label "212314".

The virtual private line fast switching process is as follows:

the process of virtual private line fast switching is shown in fig. 12. The figure only shows the forward and backward homologies of the virtual private line. The lower diagram is the state before switching, and the user accesses the Overlay network through the user CPE equipment A and connects with the opposite terminal through the user CPE equipment D. The red dotted line of A-B-D represents the main virtual private line, the green dotted line of A-C-D represents the standby virtual private line, the MPLS label value of the main virtual private line is 3, and the MPLS label value of the standby virtual private line is 4. The user data flow is packaged with a life ring layer Label of MPLS Label 1 at the forwarding node A, and then packaged with an MPLS Label 3 of the main virtual special line, and forwarded in the main virtual special line according to the Label No. 3. Each virtual private line also corresponds to a probe flow, and is also encapsulated with a corresponding MPLS label. In the figure, the main detection flow encapsulates the label No. 3, and the standby detection flow encapsulates the label No. 4.

When the active probe detects that the active virtual private line does not meet the requirement, the active probe switches to the standby virtual private line, as shown in fig. 13.

It can be seen that, at the forwarding node a, the MPLS label encapsulation of the user data flow changes from the label No. 3 to the label No. 4, and the probe flow for the active virtual private line is cancelled. And the user data flow is encapsulated with a Label MPLS Label 4 of the standby virtual private line, the user data flow is forwarded according to the standby virtual private line, and the green standby path becomes a new main path at the moment. In the whole switching process, only the flow table of the encapsulation label is changed, and the flow table forwarded by the matching label has no change. The distributed agent will report the virtual private line not meeting the condition to the centralized controller, wait for the centralized controller to issue the cancel instruction, and supplement the spare virtual private line.

The invention has the beneficial effects that:

Table 1 is a virtual private line tag management table.

TABLE 1

Claims

1. A method for forming a formation satellite routing based on an SDN framework and adopting an SR routing protocol is provided, and the method is directed to a system comprising: user terminal, satellite, gateway station, operation control center; the user terminal is communicated with the satellite, the satellite is communicated with the gateway station, and the gateway station is communicated with the operation and control center; grouping the satellites according to different orbital planes, configuring a gateway station with a controller in each group, and exchanging information between the gateway station of the controller and all the satellites on the orbital planes to finish generating a flow table; the controller gateway stations communicate through a ground link, the obtained network information is forwarded to other controller gateway stations, and each controller gateway station has full network information; the satellite, the gateway station and the user terminal support an OpenFlow protocol and are provided with southbound interfaces capable of communicating with the controller; supporting an Of-Config protocol for configuring and managing the network device; the method has the advantages that safe transmission between the controller gateway station and the adjacent transmission nodes is guaranteed, the flow table can be received from the controller gateway station, and network information including link states and resource utilization conditions is collected by forwarding data packets according to the flow table;

the routing method comprises the following steps:

step 6: completing service packet transmission;

and 7: waiting for the next task;

2. A fast rerouting method based on the routing method according to claim 1, the method comprising: