A network resource scheduling method based on a binary moth flame optimization algorithm

Abstract

This application relates to the field of network switching technology, and in particular to a network resource scheduling method based on a binary moth optimization algorithm. The method includes the following steps: generating a service requirement model, initializing the number of moths, and defining a service fitness function; calculating an initial moth path matrix and an initial flame matrix; iteratively optimizing the initial moth path matrix and flame matrix from the initial moth path matrix and initial flame matrix until a preset iteration termination condition is met, wherein in each iteration, the service fitness matrix of the current round of moths and flames is calculated and flame attenuation is performed; updating the flame positions in the flame matrix based on the updated moth paths; and selecting the optimal path from the final flame matrix as the service forwarding path. This invention solves the problem that existing technologies cannot cope with complex and dynamically changing QoS requirements, and has the characteristics of being data-independent and having low computational resource requirements.

Description

Technical Field

[0001] This application relates to the field of network switching technology, and in particular to a network resource scheduling method based on the binary moth optimization algorithm. Background Technology

[0002] With the rapid development of network technology, network scale is growing larger and larger, and business demands are becoming more diversified and dynamic. Network resource scheduling, as a core technology to ensure network quality of service (QoS), aims to find and allocate the optimal forwarding path for various services to meet their performance requirements in terms of latency, bandwidth, packet loss rate, etc.

[0003] Currently, network resource scheduling methods are mainly divided into two categories: traditional scheduling methods and intelligent scheduling methods. Traditional network resource scheduling methods, such as those based on shortest path algorithms (OSPF, IS-IS) or fixed-policy routing, are simple to implement and highly reliable, but their drawbacks are becoming increasingly apparent. On the one hand, with the rapid increase in the number of network nodes, the time spent by traditional algorithms to calculate the optimal path increases exponentially, making it difficult to meet the real-time requirements of large-scale networks. On the other hand, traditional methods are usually based on fixed routing policies, making it difficult to flexibly cope with complex and dynamically changing QoS requirements. When it is necessary to balance multiple QoS indicators (e.g., simultaneously requiring low latency and high bandwidth), they often fall short.

[0004] To overcome the limitations of traditional methods, researchers have introduced various intelligent algorithms, such as swarm intelligence algorithms and machine learning, to improve the flexibility of resource scheduling and its adaptability to large-scale networks. However, existing intelligent network resource scheduling methods generally suffer from the following drawbacks: First, most methods (especially supervised and semi-supervised learning) heavily rely on large amounts of accurate historical network data for model training, but in many real-world network environments, data collection itself is a major challenge, making the algorithms difficult to apply. Second, some intelligent algorithms are computationally intensive, requiring significant computing resources. Finally, some algorithms have poor robustness, with the risk of slow convergence or even failure to guarantee convergence, which may lead to computational failures and affect the stability of network services. Summary of the Invention

[0005] Therefore, it is necessary to provide a network resource scheduling method based on the binary moth optimization algorithm to address the above-mentioned technical problems. This method has the characteristics of being data-independent and having low computational resource requirements.

[0006] To achieve the above-mentioned objectives of this invention, the technical solution adopted is as follows:

[0007] A network resource scheduling method based on the binary moth optimization algorithm, applied to a network resource scheduling system with a software-defined networking (SDN) hybrid control architecture, includes the following steps:

[0008] Generate a business requirement model, initialize the number of moths, and define a business fitness function;

[0009] Based on forward and reverse search methods, the initial moth path matrix and the initial flame matrix are calculated.

[0010] Starting with the initial moth path matrix and initial flame matrix, the moth path matrix and flame matrix are iteratively optimized until a preset iteration termination condition is met. Each iteration calculates the current round's moth and flame fitness matrix and performs flame decay, and includes the following steps:

[0011] i. Update the moth path: Optimize the moth path using a binary search method to form a new moth path;

[0012] ii. Update flame positions: Based on the updated moth path, update the flame positions in the flame matrix;

[0013] After the iteration terminates, the optimal path is selected from the final flame matrix based on the service adaptability as the service forwarding path.

[0014] Preferably, the method is applied to a network resource scheduling system with a software-defined networking (SDN) hybrid control architecture; the network resource scheduling system includes several nodes, with the destination node as the starting node. The source node is the terminal node. The neighbor sequence number of the destination node is Specifically, it operates on the control plane of the network resource scheduling system, and the generated service forwarding paths are sent to the forwarding plane of the network resource scheduling system.

[0015] Furthermore, before generating the business requirement model, an initialization setting step is performed, including: setting the number of iterations. Divide depth coefficient Maximum path length Business adaptability upper limit And assign the maximum path length to TTL.

[0016] Furthermore, before initializing the path population, a business requirement model is generated. The specific steps are as follows:

[0017] Based on the source and destination nodes of the business, the number of moths is generated using the following formula. :

[0018]

[0019] in, For nodes Reachable nodes The number of neighbors, i.e. This represents the number of reachable neighbors of the source node. These represent the number of reachable neighbors of the source node for the destination node;

[0020] Based on the service's requirements for latency, bandwidth, packet loss rate, and jitter, define the service fitness function. for:

[0021]

[0022] in, , , and These are the latency coefficient, bandwidth coefficient, packet loss coefficient, and jitter coefficient, respectively, assigned according to the corresponding demand level. , , and These are the business paths. The delay function, bandwidth function, packet loss function, and jitter function are defined as follows:

[0023]

[0024]

[0025]

[0026]

[0027] in, , , and All are coefficients greater than 0, used to adjust the magnitudes of the delay function, bandwidth function, packet loss function, and jitter function, respectively. , , and Paths The Middle The latency, bandwidth, packet loss rate, and jitter of each node. For path The total number of nodes in the middle.

[0028] Furthermore, based on forward and reverse search methods, the initial moth path matrix and the initial flame matrix are calculated. The specific steps are as follows:

[0029] Starting from the source node of the service and the destination node, the first part of the path is generated by querying the routing table of the subsequent nodes to obtain the next hop to the destination node.

[0030] Starting from the destination node and ending at the source node, the second part of the path is generated from each reachable neighbor of the destination node in the same way as the forward search.

[0031] Reverse the order of the non-zero node elements in the second part of the path generated by the reverse search;

[0032] The first part of the path generated by the forward search is merged with the second part of the path after being reversed to form the initial moth path matrix. And set the initial flame matrix. .

[0033] Furthermore, in each iteration, the fitness matrix for the moth and flame in the current round is calculated. The specific steps are as follows:

[0034] Traversing the first k The moth path matrix of the next iteration For each path in the process, determine the validity of each path. If the starting node of the path is the business source node... Furthermore, nodes ending in non-zero are the business target nodes. If the path is valid, it is considered a valid path; otherwise, it is considered an invalid path.

[0035] For valid paths, the fitness value is calculated according to the aforementioned business fitness function. 'm' represents the sequence number, and this value is stored in the corresponding position of the moth operational fitness matrix. For invalid paths, a preset worst-fit value is assigned to them. ;

[0036] After completing the fitness calculation for all moths, reset the index m. Using the same method, divide the flame positions in the flame matrix into valid and invalid positions. For valid positions, calculate their... And store this value in the corresponding position of the flame service fitness matrix. For invalid paths, assign them an invalid path. .

[0037] Furthermore, the step of updating the moth's path specifically includes:

[0038] Moth's path Divide into sub-paths and ;

[0039] Calculate the service fitness of the two sub-paths respectively and ;

[0040] Compare and Identify the sub-paths with poor fitness and record their first and last nodes;

[0041] Search for an optimal new sub-path between the first and last nodes. ;

[0042] Using the above The new sub-path replaces the sub-path with poor adaptability in the original path, thus completing the path reorganization.

[0043] Furthermore, the flame attenuation specifically refers to:

[0044]

[0045] in As an up-rounding function, it decays starting from the flame with the worst business fitness based on the number of decaying flames, and sets the corresponding elements of the flame matrix and flame fitness matrix to 0.

[0046] Furthermore, based on the updated moth path and combined with the flame attenuation rules, the flame positions in the flame matrix are updated. The specific steps are as follows:

[0047] a. Initialize the index m, and iterate through all current flame positions;

[0048] b. Check the current m-th flame in the previous iteration. Is it 0? If If the flame is extinguished, proceed to step c; otherwise, proceed to step d.

[0049] c. Comparison And in this iteration ,if If the position of the m-th flame in this round is updated, its business adaptability is updated; otherwise, the flame position is not updated. Return to step b, and continue until all the flames have been traversed;

[0050] d. Record the index of the moth with the best business adaptability among all the current moths as Its corresponding business fitness matrix is ;like Then update the position of the flame with the optimal moth path and its business fitness; otherwise, do not update the flame position. Return to step b, and continue until all the flames have been traversed.

[0051] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

[0052] The beneficial effects of this invention are as follows:

[0053] This invention proposes a network resource scheduling method based on the binary search moth optimization algorithm, applied to a network resource scheduling system with an SDN hybrid control architecture. The method combines binary search and the moth optimization algorithm. It calculates the initial moth path matrix and initial flame matrix using forward and reverse search methods to obtain the initial path. Then, mimicking the process of a moth being attracted by a flame, it optimizes the path in the direction of optimizing service adaptability, obtaining a service forwarding path that meets the requirements. This invention is applicable to network systems of different sizes where working data is difficult to collect. It has less limitation on the number of nodes, solves the problem that existing technologies cannot handle complex and dynamically changing QoS requirements, and is suitable for network systems where working data is difficult to collect. It does not rely on large amounts of accurate historical data, has low computational resource requirements, and strong robustness. It controls the iteration convergence by selecting the number of iterations, eliminating the possibility of non-convergence. Attached Figure Description

[0054] Figure 1 This is a flowchart of a network resource scheduling method based on the binary moth optimization algorithm of the present invention;

[0055] Figure 2 This is a flowchart illustrating the network resource scheduling method based on the binary moth optimization algorithm of the present invention in Example 2.

[0056] Figure 3 This is a flowchart illustrating the specific algorithm for generating the business requirement model in Example 2;

[0057] Figure 4 This is a flowchart illustrating the specific algorithm used in Example 2 to calculate the initial moth path matrix and the initial flame matrix.

[0058] Figure 5 This is a flowchart illustrating the specific algorithm for calculating the fitness matrix of moths and flames in the current round, as shown in Example 2.

[0059] Figure 6 The flowchart of the specific algorithm in Example 2 is shown below.

[0060] Figure 7 The flowchart shows the specific algorithm in Example 2.

[0061] Figure 8 This is a schematic diagram of the network resource scheduling system in Example 3. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0063] Example 1

[0064] like Figure 1 As shown, a network resource scheduling method based on the binary moth optimization algorithm is proposed. This method is applied to a network resource scheduling system with a software-defined networking (SDN) hybrid control architecture and includes the following steps:

[0065] Generate a business requirement model, initialize the number of moths, and define a business fitness function;

[0066] Based on forward and reverse search methods, the initial moth path matrix and the initial flame matrix are calculated.

[0067] Starting with the initial moth path matrix and initial flame matrix, the moth path matrix and flame matrix are iteratively optimized until a preset iteration termination condition is met. Each iteration calculates the current round's moth and flame fitness matrix and performs flame decay, and includes the following steps:

[0068] i. Update the moth path: Optimize the moth path using a binary search method to form a new moth path;

[0069] ii. Update flame positions: Based on the updated moth path, update the flame positions in the flame matrix;

[0070] After the iteration terminates, the optimal path is selected from the final flame matrix based on the service adaptability as the service forwarding path.

[0071] Example 2

[0072] like Figure 2 As shown in this embodiment, a network resource scheduling method based on the binary moth optimization algorithm is applied to a network resource scheduling system with a software-defined networking (SDN) hybrid control architecture, and includes the following steps:

[0073] Step A: System Initialization:

[0074] Select the number of iterations Divide depth coefficient Maximum path length Business adaptability upper limit The maximum path length is assigned the value of TTL;

[0075] Step B: Generation of business requirement model;

[0076] Step C: Destination node reachability determination:

[0077] if If the destination node is unreachable, proceed to step B;

[0078] Step D: Iterative Initialization:

[0079] The initial moth path matrix was calculated based on forward and backward search methods. and the initial flame matrix ;

[0080] Step E: Update the fitness matrix for moth operations and the fitness matrix for flame operations;

[0081] Step F: If , Otherwise, proceed to step J;

[0082] Step G: Perform flame attenuation:

[0083] The number of flames decreases to ,in As an up-rounding function, based on the number of flames decaying, the decay starts from the flame with the worst business fitness and sets the corresponding elements of the flame matrix and flame fitness matrix to 0.

[0084] Step H: Update the moth path, i.e., update the moth path matrix and the moth business fitness matrix;

[0085] Step I: Update the flame position, i.e. update the flame matrix, proceed to Step E;

[0086] Step J: Iteration complete. The current business path search is complete. Optimal path selection is performed. Based on the flame matrix and the flame business fitness matrix, the flame with the best business fitness is selected, and its path is the optimal path.

[0087] Step K: Deploy the path to the forwarding plane.

[0088] like Figure 3 As shown, in a specific embodiment, the specific process for generating a business requirement model is as follows: Step B1: Read the source node of the business. and destination node ;

[0089] Step B2: Generate the number of moths The rules are as follows:

[0090]

[0091] in, For nodes Reachable nodes The number of neighbors, i.e. This represents the number of reachable neighbors of the source node. These represent the number of reachable neighbors of the source node for the destination node;

[0092] Step B3: Read the business requirements, including the requirement levels such as latency, bandwidth, packet loss rate, and jitter, which are divided into high requirement, medium requirement, and low requirement;

[0093] Step B4: Generate business adaptability functions based on the various demand levels:

[0094]

[0095] in, , , and These are the latency coefficient, bandwidth coefficient, packet loss coefficient, and jitter coefficient, respectively. The above coefficients are assigned values ​​according to the corresponding demand level: 2 for high demand, 1 for medium demand, and 0 for low demand.

[0096] , , and These are the business paths. The delay function, bandwidth function, packet loss function, and jitter function are defined as follows:

[0097]

[0098]

[0099]

[0100]

[0101] in, , , and All are coefficients greater than 0, used to adjust the magnitudes of the delay function, bandwidth function, packet loss function, and jitter function, respectively. , , and Paths The Middle The latency, bandwidth, packet loss rate, and jitter of each node. For path The total number of nodes in the middle.

[0102] like Figure 4As shown, in this embodiment, the specific process for calculating the initial moth path matrix and the initial flame matrix based on the forward search and reverse search methods is as follows:

[0103] Step D1: Calculate the initial value of the moth path matrix and the current moth index. ;

[0104] Step D2: Moth The path passes through points with index numbers of , Starting from the source node The destination node is the termination node. The neighbor sequence number of the source node is , ;

[0105] Step D3: When Proceed to step 4, when If yes, proceed to step D6; otherwise, proceed to step D9.

[0106] Step D4: Submatrix of the moth path matrix The The elements are ;

[0107] Step D5: When , If yes, proceed to step D3; otherwise, proceed to step D14.

[0108] Step D6: Read the starting node No. Check the routing tables of each neighbor to determine its connection to the termination node. The reachability of a terminal node. If yes, proceed to step D7; otherwise, proceed to step D8.

[0109] Step D7: Submatrix of the moth's path matrix The The elements are , The first node is the first One neighbor, proceed to step D5;

[0110] Step D8: Node Unreachable Termination Node , Proceed to step D6;

[0111] Step D9: Read the node The routing table is used to find the path from the node to the termination node. The next jump;

[0112] Step D10: If the next jump is not empty, proceed to step D11; otherwise, proceed to step D13.

[0113] Step D11: Use the next hop as a submatrix of the moth's path matrix. The One element;

[0114] Step D12: If If the current path search is complete, proceed to step D14; otherwise, proceed to step D5.

[0115] Step D13: Submatrix of the moth path matrix The Each element is 0;

[0116] Step D14: ,if If yes, proceed to step D15; otherwise, proceed to step D16.

[0117] Step D15: , Proceed to step 3;

[0118] Step D16: If If yes, proceed to step D17; otherwise, proceed to step D18.

[0119] Step D17: Start with the destination node The source node is the terminal node. The neighbor sequence number of the destination node is , Proceed to step D3;

[0120] Step D18: Convert the moth path matrix The Arrive at the The non-zero parts of the elements in each row are reversed, that is... , ... The non-zero parts are reversed to obtain the initial value of the moth path matrix. ;

[0121] Step D19: Initial Flame Matrix .

[0122] like Figure 5 As shown in this embodiment, the specific algorithm for calculating the fitness matrix of moths and flames in the current round is as follows:

[0123] Step E1: Calculate the business fitness of the moth. ;

[0124] Step E2: Path Sequence Number ;

[0125] Step E3: Read the first point of the path ,if If yes, proceed to step E4; otherwise, proceed to step E13.

[0126] Step E4: Read the last non-zero point in the path. ,in For matrix No. The index of the last non-zero element in the row, assuming If yes, proceed to step E5; otherwise, proceed to step E13.

[0127] Step E5: Calculate the business fitness function according to the following formula:

[0128]

[0129] Step E6: If the current calculation is for the moth's fitness, proceed to step E7; otherwise, calculate the current calculation is for the flame's fitness, proceed to step E10.

[0130] Step E7: Business adaptability of moths ,in For the first The first iteration of the moth business fitness matrix One element;

[0131] Step E8: If and , If yes, proceed to step E3; otherwise, proceed to step E9.

[0132] Step E9: If If yes, proceed to step E11; otherwise, proceed to step E14.

[0133] Step E10: The Business adaptability of individual flames ,in For the first The first iteration of the flame service fitness matrix For each element, proceed to step E8;

[0134] Step E11: Calculate the service fitness of the flame. Proceed to step E2;

[0135] Step E12: If Proceed to step E8; otherwise, proceed to step E9. Business adaptability of individual flames ,in No. The first iteration of the flame service fitness matrix For each element, proceed to step E8;

[0136] Step E13: Proceed to step E6;

[0137] Step E14: Complete the calculation and obtain the result. Sub-iteration moth business fitness matrix And the business adaptability matrix of flame .

[0138] like Figure 6 As shown, in this embodiment, the specific algorithm flow for updating the moth's path is as follows:

[0139] Step H1: Moth serial number ;

[0140] Step H2: Number of binary search operations , ,in For matrix No. The index of the last non-zero element in the row;

[0141] Step H3: Bisect the path. , , , ;

[0142] Step H4: Calculate the path according to the following formula and Business fitness function:

[0143]

[0144] Step H5: If If yes, proceed to step H6; otherwise, proceed to step H7.

[0145] Step H6: Read the path The first node and the last node Proceed to step H8;

[0146] Step H7: Read the path The first node and the last node ;

[0147] Step H8: Obtain slave nodes based on topology information. To the node All lengths not exceeding path set ,in For the node To the node All lengths not exceeding The path, This represents the maximum number of paths in the path set.

[0148] Step H9: From the path set Select the sub-path with the best business adaptability. :

[0149]

[0150] And Update to subpath Length;

[0151] Step H10: If If yes, proceed to step H11; otherwise, proceed to step H12.

[0152] Step H11: And according to the new renew Proceed to step H13;

[0153] Step H12: and according to the new renew Proceed to step H13;

[0154] Step H13: If If yes, proceed to step H14; otherwise, proceed to step H16.

[0155] Step H14: If If yes, proceed to step H15; otherwise, proceed to step H16.

[0156] Step H15: ,according to Path length update Proceed to step H3;

[0157] Step H16: The The moth's path has been updated. If , Proceed to step H2; otherwise, the paths of all moths have been updated.

[0158] Step H17: Update the moth business fitness matrix based on the new moth path matrix. .

[0159] like Figure 7As shown, in this embodiment, the specific algorithm for updating the flame position is as follows:

[0160] Step I1: Flame Sequence Number Flame attenuation sequence number ;

[0161] Step I2: Read the first Business adaptability of individual flames ,if If yes, proceed to step 6; otherwise, proceed to step 3.

[0162] Step I3: If If yes, proceed to step 4; otherwise, proceed to step 7.

[0163] Step I4: Compare with the first The second iteration Business adaptability of individual flames and the The iteration of the ... Business adaptability of moths ,if If yes, proceed to step 5; otherwise, proceed to step I6.

[0164] Step I5: Update the The location of each flame and its operational suitability:

[0165]

[0166]

[0167] Step I6: If , Proceed to step I2; otherwise, all flame positions and business adaptability updates are complete.

[0168] Step I7: Read the first... The operational fitness of a single moth is used to obtain the serial number of the moth with the best operational fitness. ;

[0169] Step I8: Compare with the first The second iteration Business adaptability of individual flames and the The iteration of the ... Business adaptability of moths ,if If yes, proceed to step I9; otherwise, proceed to step I6.

[0170] Step I9: Update the first The location of each flame and its operational suitability:

[0171]

[0172]

[0173] Proceed to step I6.

[0174] Example 3

[0175] In this embodiment, the network resource scheduling method based on the binary moth optimization algorithm is used to perform network resource scheduling on a specific network resource scheduling system.

[0176] like Figure 8 As shown, the network system in this embodiment includes 8 nodes, and the source node of the current service is... The destination node is According to the topology, the source node has 3 neighbors that can reach the destination node. If the number of neighbors that the destination node can reach the source node is 3, then ,therefore, The current business has high latency requirements, high bandwidth requirements, and low requirements for packet loss and jitter. , , and The business fitness function is:

[0177]

[0178] In this embodiment, during the forward search, the source node is The destination node is ,because If there are three neighbors, then each will receive money from the neighbor's house. , and route acquisition to The paths obtained are as follows: , , In the maximum path length Under the premise that:

[0179]

[0180]

[0181]

[0182] During reverse search, the source node is The destination node is ,because If there are three neighbors, then each will receive money from the neighbor's house. , and route acquisition to The paths obtained are as follows: , , In the maximum path length Under the premise that:

[0183]

[0184]

[0185]

[0186] Transposing the non-zero part, we get:

[0187]

[0188]

[0189]

[0190] The initial moth path matrix and the initial flame matrix are then:

[0191]

[0192] At this point, the initialization step of the algorithm in this embodiment is complete. Next, the system will calculate the initial fitness based on this initial moth and flame matrix, and then begin the iterative optimization process until the optimal path is found and deployed.

[0193] Example 4

[0194] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

Claims

1. A network resource scheduling method based on the binary moth optimization algorithm, characterized in that, Includes the following steps: Generate a business requirement model, initialize the number of moths, and define a business fitness function; Based on forward and reverse search methods, the initial moth path matrix and initial flame matrix are calculated. The specific steps are as follows: Starting from the source node of the service and the destination node, the first part of the path is generated by querying the routing table of the subsequent nodes to obtain the next hop to the destination node. Starting from the destination node and ending at the source node, the second part of the path is generated from each reachable neighbor of the destination node in the same way as the forward search. Reverse the order of the non-zero node elements in the second part of the path generated by the reverse search; The first part of the path generated by the forward search is merged with the second part of the path after being reversed to form the initial moth path matrix. And set the initial flame matrix. Starting with the initial moth path matrix and initial flame matrix, the moth path matrix and flame matrix are iteratively optimized until a preset iteration termination condition is met. Each iteration calculates the current round's moth and flame fitness matrix and performs flame decay, and includes the following steps: i. Update the moth path: Optimize the moth path using a binary search method to form a new moth path; ii. Update flame positions: Based on the updated moth path, update the flame positions in the flame matrix; In each iteration, the fitness matrix for the moth and flame in the current round is calculated. The specific steps are as follows: Traversing the first k The moth path matrix of the next iteration For each path in the process, determine the validity of each path. If the starting node of the path is the business source node... Furthermore, nodes ending in non-zero are the business target nodes. If the path is valid, it is considered a valid path; otherwise, it is considered an invalid path. For valid paths, the fitness value is calculated according to the aforementioned business fitness function. 'm' represents the sequence number, and this value is stored in the corresponding position of the moth operational fitness matrix. For invalid paths, a preset worst-fit value is assigned to them. ; After completing the fitness calculation for all moths, reset the index m. Using the same method, divide the flame positions in the flame matrix into valid and invalid positions. For valid positions, calculate their... The fitness value of the effective position is then stored in the corresponding position of the flame service fitness matrix. For invalid paths, assign them an invalid path. ; After the iteration terminates, the optimal path is selected from the final flame matrix based on the service adaptability as the service forwarding path.

2. The method according to claim 1, characterized in that, The method described above is applied to a network resource scheduling system with a software-defined networking (SDN) hybrid control architecture; the network resource scheduling system includes several nodes, with the destination node as the starting node. The source node is the terminal node. The neighbor sequence number of the destination node is ; Specifically, it operates on the control plane of the network resource scheduling system, and the generated service forwarding paths are sent to the forwarding plane of the network resource scheduling system.

3. The method according to claim 2, characterized in that, Before generating the business requirement model, an initialization setting step is also performed, including: setting the number of iterations. Divide depth coefficient Maximum path length Business adaptability upper limit And assign the maximum path length to TTL.

4. The method according to claim 3, characterized in that, Before initializing the path population, a business requirement model is generated. The specific steps are as follows: Based on the source and destination nodes of the business, the number of moths is generated using the following formula. : in, For nodes Reachable nodes The number of neighbors, i.e. This represents the number of reachable neighbors of the source node. This represents the number of reachable neighbors of the source node for the destination node. Based on the service's requirements for latency, bandwidth, packet loss rate, and jitter, define the service fitness function. for: in, , , and These are the latency coefficient, bandwidth coefficient, packet loss coefficient, and jitter coefficient, respectively, assigned according to the corresponding demand level. , , and These are the business paths. The delay function, bandwidth function, packet loss function, and jitter function are defined as follows: in, , , and All are coefficients greater than 0, used to adjust the magnitudes of the delay function, bandwidth function, packet loss function, and jitter function, respectively. , , and Paths The Middle The latency, bandwidth, packet loss rate, and jitter of each node. For path The total number of nodes in the middle.

5. The method according to claim 1, characterized in that, The specific steps for updating the moth's path are as follows: Moth's path Divide into sub-paths and ; Calculate the service fitness of the two sub-paths respectively. and ; Compare and Identify the sub-paths with poor fitness and record their first and last nodes; Search for an optimal new sub-path between the first and last nodes. ; Using the above The new sub-path replaces the sub-path with poor adaptability in the original path, thus completing the path reorganization.

6. The method according to claim 4, characterized in that, The flame attenuation mentioned above specifically refers to: in As an up-rounding function, it decays starting from the flame with the worst business fitness based on the number of decaying flames, and sets the corresponding elements of the flame matrix and flame fitness matrix to 0.

7. The method according to claim 6, characterized in that, Based on the updated moth path and the flame attenuation rule, the flame position in the flame matrix is ​​updated. The specific steps are as follows: a. Initialize the index m, and iterate through all current flame positions; b. Check the current m-th flame in the previous iteration. Is it 0? If If the flame is extinguished, proceed to step c; otherwise, proceed to step d. c. Comparison And in this iteration ,if If the position of the m-th flame in this round is updated, its business adaptability is updated; otherwise, the flame position is not updated. Return to step b, and continue until all the flames have been traversed; d. Record the index of the moth with the best business adaptability among all the current moths as Its corresponding business fitness matrix is ;like Then update the position of the flame with the optimal moth path and its business fitness; otherwise, do not update the flame position. Return to step b, and continue until all the flames have been traversed.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images