Cross-domain multi-modal network-oriented modal service quality guarantee method

Abstract

The invention relates to the technical field of multi-modal networks, and discloses a cross-domain multi-modal network-oriented modal service quality assurance method, which comprises the following steps of: sensing a network congestion state of a backbone network by using a simple network management protocol (SNMP), calculating an instantaneous bandwidth and a data packet average delay of each network switching node in the backbone network, and calculating the network congestion state of each network switching node in the backbone network; therefore, a complete network state diagram of the backbone network is constructed; generating different forwarding paths based on the network state diagram of the backbone network; the cross-domain modal data packets of the corresponding multi-modal network domains are packaged through the co-processors; and taking different forwarding paths as forwarding channels of different encapsulated cross-domain modal data packets to complete modal traffic cross-domain channel mapping. According to the invention, on the basis of network state perception, the multi-modal network cross-domain interconnection for guaranteeing the service quality is realized through presetting the forwarding channel.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-modal networks, and in particular, to a method for ensuring the quality of modal services for cross-domain multi-modal networks. Background Art

[0002] A multi-modal network is a network development paradigm in which the network technology regime is separated from the physical platform, aiming to efficiently carry diversified services on network element devices with a unified architecture and provide multi-service convergence services using the same network infrastructure. Through the full-dimensional definable technology, the multi-modal network seeks the intersection of service quality commitment (S), resource reusability (M), and service diversity (V) to achieve the ideal network vision.

[0003] This network dynamically loads and runs various network technology regimes in the form of modalities in a multi-modal network environment, so as to meet the user's needs of "loading the corresponding network modality according to what network service the application needs". The multi-modal network processes according to the message format, routing protocol, switching method, forwarding logic, etc. defined by the modality, realizes the symbiotic coexistence, dynamic evolution, and transformative development of multiple modalities on the same physical network platform, thereby supporting the "on-demand customization" of network structure and attributes such as function, performance, efficiency, and security, and constructing an evolvable and diversified network technology regime ecosystem.

[0004] Through software-defined interconnection, full-dimensional definable technology, etc., the multi-modal network breaks through the single-bearing structure development mode of traditional networks, forms a diversified network development paradigm for diversified vertical industry needs, and realizes the integrated development that supports both existing technology regimes, evolutionary technology regimes, and transformative technology regimes at the same time. Based on the multi-modal network environment, the network will develop into the main components of computing, storage, and services, and become the direct actor and assembler for global resource scheduling and service provision.

[0005] Service traffic is forwarded in the multi-modal network domain through different network protocol stacks in multi-modal network element devices. However, for cross-domain multi-modal networks, network traffic needs to be forwarded through an IP-based backbone network between different multi-modal network domains. Existing message encapsulation and path selection methods are difficult to achieve interconnection across traditional IP networks, and multi-modal networks with cross-domain networking are difficult to provide service quality commitments to modal users. Specifically, it is impossible to efficiently encapsulate different network service data packets so that they can be interconnected on the Internet; it is impossible to promise the bandwidth of cross-domain forwarded traffic to network modal users; it is impossible to promise the delay of cross-domain forwarded traffic to network modal users, etc.

[0006] How to provide a network modal service quality guarantee scheme for cross-domain multi-modal networks is a research hotspot in the current academic and industrial circles.

[0007] Packet encapsulation based on IP tunnels and shortest path first: Simply add a fixed IP header at the front end of multi-modal network packets for encapsulation. Through configuring a coprocessor and using network address translation technology, this encapsulation scheme can achieve interconnection between multi-modal network domains. The backbone network devices generate the shortest path for network traffic through OSPF. That is, let the multi-modal network packets pass through the IP tunnel to achieve interconnection between different network domains. The packet encapsulation based on IP tunnels and shortest path first simply adds an IP header to achieve cross-domain interconnection, and still uses the "best effort" mechanism in the IP network transmission, thus unable to provide cross-domain service quality guarantee for different network modalities. Packets of different network modalities are still forwarded on the backbone network by the traditional IP protocol mechanism, and it is difficult to achieve classified cross-domain performance guarantee and modality isolation on the backbone network according to different network modality identifiers, and to realize the dynamic allocation and utilization of network resources.

[0008] Packet encapsulation based on source routing and specified path forwarding: By adding a source routing header at the front end of multi-modal network packets, use the coprocessor to determine the forwarding path of the network packet and encapsulate it into the packet header to achieve a custom packet forwarding path. By avoiding congested nodes, the service quality of network services can be effectively improved. The existing packet encapsulation based on source routing and specified path forwarding using SRv6 cannot achieve fine-grained traffic path generation with environmental awareness. That is, SRv6 can achieve "best effort" traffic optimization, but cannot provide targeted service quality commitments and modality traffic isolation for different network modality users in multi-modal networks.

[0009] From the above analysis, it can be seen that the existing technologies can achieve cross-domain interconnection of multi-modal network domains, but it is difficult to make service quality commitments for different modality services for cross-domain traffic, resulting in cross-domain traffic still enjoying "best effort" services rather than personalized customization capabilities in multi-modal networks. Summary of the Invention

[0010] In view of the above problems, the present invention proposes a method for guaranteeing modality service quality for cross-domain multi-modal networks, which can encapsulate multi-modal network traffic to achieve cross-domain interconnection between different network domains. At the same time, by using real-time traffic status awareness technology and congestion avoidance routing technology, it realizes service quality guarantee for modality traffic at different levels. The present invention uses network congestion awareness and traffic engineering to evenly utilize the idle nodes and idle bandwidth of the IP network. A source routing-based, fine-grained cross-domain traffic routing technology is proposed to generate customized and dynamic cross-domain interconnection forwarding paths for each network modality, so as to achieve the maximum service quality commitment and modality isolation ability.

[0011] To achieve the above object, the present invention adopts the following technical solutions:

[0012] A method for ensuring the quality of modal services for a cross - domain multimodal network, where the multimodal network includes multiple multimodal network domains, and the multiple multimodal network domains are interconnected through a backbone network. Each multimodal network domain contains a coprocessor, and the method includes:

[0013] Using the Simple Network Management Protocol (SNMP) to sense the network congestion status of the backbone network, calculating the instantaneous bandwidth and average packet delay of each network switching node in the backbone network, so as to construct a complete backbone network status graph;

[0014] Generating different forwarding paths based on the backbone network status graph;

[0015] Encapsulating cross - domain modal data packets of the corresponding multimodal network domain through each coprocessor;

[0016] Using different forwarding paths as the forwarding channels for the encapsulated different cross - domain modal data packets to complete the mapping of modal traffic cross - domain channels.

[0017] Further, constructing a complete backbone network status graph in the following way:

[0018] Configuring a routing protocol on the coprocessor, and constructing a backbone network topology graph G=(V, E) through a routing algorithm; where G is a directed graph, V represents network switching nodes, and E represents the connection paths between network switching nodes;

[0019] Configuring the SNMP protocol on the backbone network switching nodes, and using this protocol on the coprocessor to calculate the port instantaneous bandwidth of different network switching nodes, realizing the perception of the backbone network status, and further completing the construction of the backbone network status graph.

[0020] Further, in the backbone network topology graph, if there are multiple hard - isolation pipelines between two network switching nodes, then construct the multiple hard - isolation pipelines as different network edges.

[0021] Further, calculating the port instantaneous bandwidth in the following way:

[0022]

[0023] Among them, InBitRate represents the in - port instantaneous bandwidth, OutBitRate represents the out - port instantaneous bandwidth, represents the value of the input data record register at time t1, represents the value of the output data record register at time t1.

[0024] Further, generating different forwarding paths based on the backbone network status graph includes:

[0025] Step a1: Add the coprocessor nodes of the multimodal network domain to the backbone network topology diagram to construct the network topology diagram G after adding the coprocessor. ′ = G ∪ V start ∪ S end ; where V start represents the coprocessor of the current multimodal network domain, and S end = {V end1 , V end2 , …, V endn} represents the set of coprocessors of other multimodal network domains except the current multimodal network domain; for each coprocessor V endi ∈ S end , construct the forwarding path set SP start from V endi to V i = {p j}, and initialize and the instantaneous bandwidth threshold b thres ;

[0026] Step a2: On G′, use the Dijkstra algorithm to calculate the shortest path p start from V endi to V j , and add it to SP i ;

[0027] Step a3: Select the minimum instantaneous bandwidth b j in all network edges e k contained in p min ;

[0028] Step a4: Subtract b k from the remaining bandwidth of e min , and update G′;

[0029] Step a5: If b min < b thres , then stop generating new forwarding channels, otherwise jump to Step a2 and repeat generating new forwarding paths.

[0030] Furthermore, the coprocessor includes a message encapsulation module and a modality recognition module;

[0031] The message encapsulation module adds a message header to the data in the multimodal network domain according to the specified mapping rule, and the modality information of the data is encoded into a specified protocol identification field and encapsulated in the new message header;

[0032] The modality recognition module converts the cross-domain modality data packet into a message header that can be forwarded in the multimodal network domain according to the protocol identification field in the header of the cross-domain modality data packet in the backbone network.

[0033] Furthermore, the process of using different forwarding paths as the forwarding channels for different cross-domain modal data packets after encapsulation to complete the cross-domain channel mapping of modal traffic includes:

[0034] Step b1: Initialize the modal and path mapping set of the traffic from V start to V endi ;

[0035] Step b2: Select a modal k ∈ M, where M is the set of network modals. Traverse SP i , and select the path set P i whose bandwidth and forwarding delay both meet the requirements of modal k. Then select the path v with the minimum channel bandwidth in P i . If P i is an empty set, it means that quality of service guarantee cannot be provided for cross-domain traffic, and jump to step b4;

[0036] Step b3: Update Map i = Map i ∪ (k, v);

[0037] Step b4: If not all modals have been traversed, jump to step b2 to traverse the next modal. If all traversals have been completed, end the algorithm.

[0038] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0039] The present invention realizes the perception of network congestion status through a coprocessor, and at the same time uses the message header to achieve specified path forwarding across multiple modal network domains. Through the designed forwarding channel generation algorithm, the generation of forwarding channels for cross-domain multi-modal networks is realized. Also, through the designed cross-domain channel mapping of modal traffic, the generation of the mapping set is realized.

[0040] Based on network status perception, the present invention realizes cross-domain interconnection of multi-modal networks with guaranteed quality of service through preset forwarding channels. Compared with hard isolation technology, it can achieve better resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a schematic flowchart of a method for guaranteeing modal quality of service for a cross-domain multi-modal network provided by an embodiment of the present invention;

[0042] Figure 2 is a schematic diagram of a cross-domain networking architecture of a multi-modal network provided by an embodiment of the present invention;

[0043] Figure 3 is a schematic diagram of the cooperation principle of the coprocessor message encapsulation module and the modal recognition module provided by an embodiment of the present invention;

[0044] Figure 4 An example of a cross - domain multi - modal network topology provided by an embodiment of the present invention;

[0045] Figure 5 An example of the updated topology G' provided by an embodiment of the present invention. Detailed implementation manners

[0046] The present invention will be further explained and illustrated below in conjunction with the accompanying drawings and specific embodiments:

[0047] As shown in Figure 1 A method for ensuring the quality of modal services for a cross - domain multi - modal network includes:

[0048] Network congestion status awareness: Use SNMP (Simple Network Management Protocol) to sense the network congestion status of the network, calculate the instantaneous bandwidth and average packet delay of each network switching node in the backbone network, so as to construct a complete backbone network status graph.

[0049] Forwarding channel generation: Generate different forwarding paths based on the backbone network status graph.

[0050] Network modal packet encapsulation: Each coprocessor encapsulates the cross - domain modal packets of the corresponding multi - modal network domain, so that the cross - domain modal packets can be transmitted within the network.

[0051] Modal traffic cross - domain channel mapping: Use different forwarding paths as the forwarding channels for different encapsulated cross - domain modal packets to complete the modal traffic cross - domain channel mapping. Different forwarding paths are used for different modalities, so as to ensure that there is no competition in the network traffic of different modalities, thereby realizing the guarantee of service quality and isolation between different modalities.

[0052] Figure 2 It is a cross - domain networking architecture for a multi - modal network. Among them, it includes multi - modal network domain 1 and multi - modal network domain 2. The two multi - modal network domains are interconnected through the backbone network. Among them, the coprocessor of the multi - modal network domain is responsible for the encapsulation and forwarding of cross - domain traffic and is directly connected to the backbone network elements.

[0053] As an implementable manner, the method specifically includes the following content.

[0054] I. Network congestion status awareness

[0055] First, configure a routing protocol on the coprocessor. Construct a network topology G=(V, E) through a routing algorithm. Among them, G is a directed graph, V represents network switching nodes, and E represents the connection paths between network switching nodes.

[0056] Further, configure the SNMP protocol on the backbone network switching nodes (switches), and use this protocol on the coprocessor to calculate the instantaneous bandwidth of the ports of different network switching nodes. The formula for the instantaneous bandwidth of a port is as follows:

[0057]

[0058] where InBitRate represents the instantaneous bandwidth of the input port, and OutBitRate is the instantaneous bandwidth of the output port, both in megabits per second. represents the value of the input data record register at time t1, represents the value of the output data record register at time t1. The coprocessor calculates the instantaneous bandwidth of the backbone switch in this way to realize the perception of the network congestion state.

[0059] II. Forwarding Channel Generation

[0060] After completing the perception of the backbone network state, different forwarding paths need to be generated based on the state perception, so as to provide forwarding channels for different modalities and realize the guaranteed quality of service occupancy of channel resources by different modality traffic.

[0061] Step S11: Initialize the network topology graph. First, add the coprocessor nodes of the multimodal network domain to the backbone topology graph. Denote the current coprocessor as V start , and the set of coprocessors in other multimodal network domains as S end = {V end1 , V end , …, V endn}. The network topology graph after adding the coprocessor is G ′ = G ∪ V start ∪ S end . For each coprocessor V endi ∈ S end , construct the set of forwarding paths SP start from the current coprocessor V endi to V i = {p j}. Initialize the set of forwarding paths and the instantaneous bandwidth threshold b thres .

[0062] Step S12: To generate SP i , the present invention uses the Dijkstra algorithm on the network topology graph G' to calculate the shortest path p start from the current coprocessor V endi to V j , and add it to the set of forwarding paths SP i , that is, SP i = SPi ∪p j . Among them, the cost of Dijkstra is the reciprocal of the link delay. That is:

[0063]

[0064] Among them, bw is the bandwidth of the link, and bitrate is the instantaneous rate of the existing bandwidth of the switch. The cost metric cost adopted by the present invention is the bandwidth delay of the switch.

[0065] Step S13: The path p generated through step S12 j ={v k |v k ∈V}. For the path p j containing the network edge e k =(v k , v k+1 ), select the minimum instantaneous bandwidth among all the edges, that is

[0066] Step S14: For all network edges e k , subtract the minimum instantaneous bandwidth b k of the path p j from the remaining bandwidth of e min . Thus, update the network topology graph G'.

[0067] Step S15: If the minimum instantaneous bandwidth b min <b thres , then stop generating new forwarding channels, otherwise jump to step S12 and repeat generating new forwarding paths.

[0068] So far, the generation of the forwarding path SP start from the coprocessor V endi to the coprocessor V i is completed. Similarly, the generation of the forwarding path SP start from the coprocessor V endn to the coprocessor V n is completed, providing a forwarding path candidate for the forwarding of cross-domain multimodal network traffic.

[0069] It should be further noted that in the backbone network topology graph, if there are multiple hard isolation pipes between two nodes, these hard isolation pipes are regarded as different network edges.

[0070] III. Network Modal Packet Encapsulation

[0071] When cross-domain packets pass through the coprocessor, message encapsulation is required to achieve traffic forwarding in the backbone network.

[0072] To ensure that cross - domain data packets are forwarded along the specified path, two main solutions can be adopted:

[0073] The first solution is to change the forwarding table in the network to specify the forwarding path of the data packet. This solution requires identifying multi - modal data packets to ensure that switching devices in the backbone network can correctly recognize multi - modal data packets, so that the data packets are forwarded along the specified path. However, this solution mainly has two relatively serious problems: one is that it is impossible to change the forwarding tables of all network switching devices. The other is that the complexity of changing the forwarding path is relatively large, involving the adjustment of multiple network switching devices, which is likely to cause inconsistencies between the data plane and the control plane of the switching devices, resulting in loops and large - scale packet loss in the network.

[0074] The second solution is to use source routing to specify the path forwarding of network data packets. The present invention uses SRv6 to specify the path forwarding of network data packets, so as to ensure that cross - domain data packets in a multi - modal network are forwarded through the specified forwarding channel, thereby achieving service quality guarantee and isolation of different modalities. In this solution, the forwarding path is encapsulated in the coprocessor, without involving changes to the forwarding tables of a large number of devices, and there will be no problem of inconsistency leading to network loop packet loss.

[0075] The coprocessor contains two modules: packet encapsulation and modality recognition. Among them, the packet encapsulation module mainly adds the packet header to the data in the multi - modal network domain according to the specified mapping rules. Its modality information will also be encoded into a specified protocol identification field and encapsulated in the new packet header. The function of the modality recognition module is mainly to convert the cross - domain modality data packet into a packet header that can be forwarded in the multi - modal network domain according to the identification field in the header of the cross - domain modality data packet in the backbone network. The working principles of the above two modules are as Figure 3 shown.

[0076] The present invention adopts a packet encapsulation format based on SRv6. To encapsulate multi - modal packets into SRv6 packets, it can be achieved by defining a new Next Header value. According to the IPv6 packet encapsulation format, the present invention uses Next Header to indicate the next extension header or payload type of the packet. The protocol number and protocol mapping relationship of Next Header are shown in Table 1.

[0077] Table 1 Next Header protocol number

[0078] Next Header value Protocol 4 IPv4 41 IPv6 43 Routing Header 200 Geo 201 NDN 202 MF

[0079] IV. Cross - domain channel mapping of modality traffic

[0080] After the generation of the cross - domain channel and the definition of the cross - domain packet header fields are completed, the present invention further completes the mapping of the cross - domain forwarding channel in the network modality domain.

[0081] Coprocessor V of the press association start To coprocessor V endi The set of forwarding paths is SP i . Denote the set of network modes as M = {m i}. Denote Map i = {(k, v)|k ∈ M, v ∈ SP i}. Map i represents the mode and path mapping set of the traffic from coprocessor V start to coprocessor V endi .

[0082] Step S21: Initialize the mapping set

[0083] Step S22: Select k ∈ M and traverse the forwarding path set p ∈ SP i . Select the path set P i whose bandwidth and forwarding delay both meet the requirements of mode k, and select the path v with the minimum channel bandwidth in the set P i . If P i is an empty set, it means that quality of service guarantee cannot be provided for cross-domain traffic. Jump to step S24

[0084] Step S23: Update the mapping set Map i = Map i ∪ (k, v).

[0085] Step S24: If not all modes have been traversed, jump to step S22 to traverse the next mode. If the traversal has been completed, end the algorithm

[0086] Similarly, the set of forwarding paths from coprocessor V start to coprocessor V endn is SP n , and the mapping set Map n is generated through the cross-domain channel mapping of the modal traffic of the present invention. In summary, the mapping of the modal traffic cross-domain channel is completed

[0087] As a specific implementable manner, as Figure 4 shown, taking the example of a coprocessor 1 in a multi-modal network domain 1 forwarding cross-domain data packets to a coprocessor 2 in a multi-modal network 2, the working process of the present invention is specifically described

[0088] First, construct a network topology graph G, and after adding two coprocessors, it becomes G'. Among them, the nodes and edges of G' are all as Figure 4 shown

[0089] Secondly, the Dijkstra algorithm is used to calculate the shortest path. According to the definition of cost, the costs from v1 to v2, from v2 to v5, and from v5 to v6 are all 0.02. The node costs from vs1 to v1 and from v6 to vs2 are 0.01. The costs from v1 to v3, from v3 to v4, and from v4 to v6 are all 0.05. Using the Dijkstra algorithm to calculate the path from vs1 to vs2, the shortest path is the path represented by Channel 1. The bandwidth of Channel 1 is 50G.

[0090] Furthermore, subtract 50G from the remaining bandwidth of the edges in the path represented by Channel 1. Thus, the topology is as Figure 5 shown.

[0091] Furthermore, calculate the path in the updated topology G′, and Channel 2 can be obtained. In summary, the establishment of the channels is completed. That is, Channel 1 and Channel 2.

[0092] After completing the channel construction, it is necessary to generate a modal mapping table. Both the existing Multimodal Network Domain 1 and Multimodal Network Domain 2 include three modalities: MF, NDN, and IPv4, and the traffic of each modality is 20G. The following is a working example of the modal mapping table:

[0093] First, select the MF modality. The available channel set includes Channel 1 and Channel 2. Since the resources of Channel 2 meet the modal requirements and the channel bandwidth is smaller than that of Channel 1, the key-value pair (MF, Channel 2) is added to the mapping set. Since the remaining bandwidth of Channel 2 is 0, Channel 2 is removed.

[0094] Furthermore, traverse the NDN modality. The available channel set includes Channel 1. Since Channel 1 meets the modal requirements, the key-value pair (NDN, Channel 1) is added to the mapping set. Since there is still remaining bandwidth in Channel 1, Channel 1 is not removed.

[0095] Finally, traverse the IPv4 modality. The available channel set includes Channel 1. Since Channel 1 meets the modal requirements, the key-value pair (IPv4, Channel 1) is added to the mapping set. Since there is still remaining bandwidth in Channel 1, Channel 1 is not removed.

[0096] In summary, the mapping between the modalities and channels is completed.

[0097] After completing the channel mapping, encapsulate the network modal traffic. The packet encapsulation is implemented using the SRv6 protocol. Add the SRv6 packet header to the NDN and IPv4 data packets, and write the node ID of Channel 1 into the SID in the SRv6 header. Add the SRv6 packet header to the MF data packet, and write the node ID of Channel 2 into the SID in the SRv6 header.

[0098] Finally, the coprocessor completes the forwarding of the modal message to the backbone network node, and the receiving coprocessor removes the SRv6 header and restores it to the message header that can be processed by the multi-modal network.

[0099] Finally, the cross-domain interconnection of the example multi-modal network message with guaranteed quality of service is completed.

[0100] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for ensuring the quality of modal service for a cross-domain multimodal network, the multimodal network comprising a plurality of multimodal network domains interconnected by a backbone network, each multimodal network domain including a coprocessor, characterized in that, Including: Using the Simple Network Management Protocol (SNMP) to sense the network congestion status of the backbone network, calculating the instantaneous bandwidth and average packet delay of each network switching node in the backbone network, so as to construct a complete backbone network status graph; Generating different forwarding paths based on the backbone network status graph; Encapsulating cross-domain modal data packets of corresponding multi-modal network domains through each coprocessor; Using different forwarding paths as the forwarding channels for different cross-domain modal data packets after encapsulation to complete the cross-domain channel mapping of modal traffic.

2. The method for ensuring the quality of modal service for a cross-domain multi-modal network according to claim 1, wherein Constructing a complete backbone network status graph in the following way: Configuring a routing protocol on the coprocessor and constructing a backbone network topology graph G=(V, E) through a routing algorithm; where G is a directed graph, V represents network switching nodes, and E represents the connection paths between network switching nodes; Configuring the SNMP protocol on the backbone network switching nodes and using this protocol on the coprocessor to calculate the port instantaneous bandwidth of different network switching nodes, realizing the perception of the backbone network status, and further completing the construction of the backbone network status graph.

3. The method for ensuring the quality of modal service for a cross-domain multi-modal network according to claim 2, characterized in that, In the backbone network topology graph, if there are multiple hard isolation pipes between two network switching nodes, then construct the multiple hard isolation pipes as different network edges.

4. A method for ensuring the quality of modal services for a cross-domain multi-modal network according to claim 2, characterized in that, Calculating the port instantaneous bandwidth in the following way: Among them, InBitRate represents the instantaneous bandwidth of the input port, and OutBitRate represents the instantaneous bandwidth of the output port. represents the value of the input data record register at time t1. represents the value of the output data record register at time t1.

5. A method for ensuring the quality of modal services for a cross-domain multi-modal network according to claim 2, characterized in that, Generating different forwarding paths based on the backbone network status graph includes: Step a1: Add the coprocessor nodes of the multimodal network domain to the backbone network topology diagram to construct the network topology diagram G after adding the coprocessor ′ = G ∪ V start ∪ S end ; where V start represents the coprocessor of the current multimodal network domain, and S end = {V end1 , V end2 , …, V endn} represents the set of coprocessors of other multimodal network domains except the current multimodal network domain; for each coprocessor V endi ∈ S end , construct the forwarding path set SP start from V endi to V i = {p j}, initialize and the instantaneous bandwidth threshold b thr ; Step a2: On G′, use the Dijkstra algorithm to calculate the shortest path p start from V endi to V j , and add it to SP i ; Step a3: Select p j for all network edges e k in it with the minimum instantaneous bandwidth b min ; Step a4: Subtract b k from the remaining bandwidth of e min and update G'; Step a5: If b ,in <b thres , stop generating new forwarding channels; otherwise, jump to step a2 and repeat generating new forwarding paths.

6. A method for ensuring the quality of modal services for a cross-domain multi-modal network according to claim 1, characterized in that The coprocessor includes a message encapsulation module and a modal recognition module; The message encapsulation module adds a message header to the data in the multi-modal network domain according to the specified mapping rule, and the modal information of the data is encoded into a specified protocol identification field and encapsulated in the new message header; The modal recognition module converts the cross-domain modal data packet into a message header that can be forwarded in the multi-modal network domain according to the protocol identification field in the header of the cross-domain modal data packet in the backbone network.

7. A method for ensuring the quality of modal service for a cross-domain multi-modal network according to claim 5, characterized in that, Using different forwarding paths as the forwarding channels for different cross-domain modal data packets after encapsulation to complete the cross-domain channel mapping of modal traffic includes: Step b1: Initialize the modal and path mapping set of the flow from V start to V endi ​ Step b2: Select a mode k ∈ M, where M is the set of network modes, and traverse SP i , and select a set of paths P whose bandwidth and forwarding delay both meet the requirements of mode k i , and select the path v with the minimum channel bandwidth in P i . If P i is an empty set, it means that quality of service guarantee cannot be provided for cross-domain traffic, and jump to step b4; Step b3: Update Map i = Map i ∪(k, v); Step b4: If not all modalities have been traversed, jump to step b2 to traverse the next modality. If all traversals have been completed, end the algorithm.

Images