A long-distance communication transmission method and system for a domestic switch

Abstract

The present invention relates to a long-distance communication transmission method and system for a domestic switch, which relates to the field of communication transmission. The method includes obtaining the packet size, the constraint duration, and the available bandwidth of multiple transmission links between the initialization and the receiving end of the receiving end at the same time, slicing the data to be transmitted to obtain a slice sequence with the same slice size; randomly grouping the slice sequence with the number of transmission links as the maximum group number constraint, respectively inputting the number of data slices, the first required bandwidth, and the constraint duration of each group into the required bandwidth matching model, and outputting the second required bandwidth; calculating the ratio of the second required bandwidth to the decision ratio threshold to obtain the minimum available bandwidth, and obtaining a data transmission scheme based on the available bandwidth for transmission link matching, which solves the problems that the bandwidth requirement after packet slicing cannot be accurately predicted during long-distance communication transmission of the switch, resulting in insufficient transmission real-time performance and adaptability, and the inability to achieve efficient and stable long-distance communication transmission.

Description

Technical Field

[0001] The present invention relates to the field of communication transmission, and in particular, to a long-distance communication transmission method and system for a domestic switch. Background Art

[0002] In long-distance communication transmission, the switch, as a core device, undertakes the important tasks of data forwarding and transmission. With the continuous development of network technology, the scale and complexity of data packets are increasing continuously, which puts higher requirements on the transmission capacity of the switch. Especially in the field of domestic switches, how to achieve efficient and stable long-distance communication transmission while ensuring data security has become a technical problem to be solved urgently at present.

[0003] During the communication process, when the data packet is too large, the switch usually divides the data packet into small slice data for transmission. However, this simple slicing method has many problems. On the one hand, the bandwidth required by the multiple sub-data packets after the data packet is sliced is not linearly reduced according to the slicing size ratio, which makes it difficult to accurately predict the bandwidth resources required by each sub-data packet during the actual transmission process. On the other hand, if relying on manual bandwidth configuration by users, not only is the efficiency low, but also it is difficult to meet the timeliness requirements of data transmission. Especially in a dynamically changing network environment, the allocation and management of bandwidth resources become more complex and difficult. Summary of the Invention

[0004] Aiming at the technical problems in the prior art that the bandwidth requirements after slicing the data packets in the long-distance communication transmission of the switch cannot be accurately predicted, resulting in insufficient transmission real-time performance and adaptability, and it is impossible to achieve efficient and stable long-distance communication transmission, the present invention provides a long-distance communication transmission method and system for a domestic switch to solve the above problems.

[0005] The technical solutions of the present invention to solve the above technical problems are as follows:

[0006] In a first aspect, the present invention provides a long-distance communication transmission method for a domestic switch, including: obtaining the packet size, constraint duration, and target receiving end of the data to be transmitted, and simultaneously initializing the available bandwidths of multiple transmission links with the target receiving end; slicing the data to be transmitted to obtain a data slice sequence, where each data slice has the same size, and wherein the data slice has a preset first required bandwidth; randomly grouping the data slice sequence with the number of transmission links as the maximum number of groups constraint to obtain several groups of data slices, and for each group of data slices, inputting the number of data slices, the first required bandwidth, and the constraint duration into a required bandwidth matching model respectively to output several second required bandwidths, the required bandwidth matching model being constructed by machine learning through multiple groups of data, and any one of the multiple groups of data includes: the number of data slice records of a preset communication protocol, the recorded bandwidth of the unit slice requirement, the constraint recorded duration, and a label indicating the output required bandwidth; calculating the ratios of the several second required bandwidths to a decision ratio threshold respectively to obtain several minimum available bandwidths, and performing transmission link matching based on the multiple available bandwidths to obtain a data transmission scheme to transmit the data to be transmitted to the target receiving end.

[0007] In a second aspect, the present invention provides a long-distance communication transmission system for a domestic switch, including: an initialization module for obtaining the packet size, constraint duration, and target receiving end of the data to be transmitted, and simultaneously initializing the available bandwidths of multiple transmission links with the target receiving end; a data slicing module for slicing the data to be transmitted to obtain a data slice sequence, where each data slice has the same size, and wherein the data slice has a preset first required bandwidth; a model matching module for randomly grouping the data slice sequence with the number of transmission links as the maximum number of groups constraint to obtain several groups of data slices, and for each group of data slices, inputting the number of data slices, the first required bandwidth, and the constraint duration into a required bandwidth matching model respectively to output several second required bandwidths, the required bandwidth matching model being constructed by machine learning through multiple groups of data, and any one of the multiple groups of data includes: the number of data slice records of a preset communication protocol, the recorded bandwidth of the unit slice requirement, the constraint recorded duration, and a label indicating the output required bandwidth; a scheme obtaining module for calculating the ratios of the several second required bandwidths to a decision ratio threshold respectively to obtain several minimum available bandwidths, and performing transmission link matching based on the multiple available bandwidths to obtain a data transmission scheme to transmit the data to be transmitted to the target receiving end.

[0008] The beneficial effects of the present invention are as follows: By slicing the data to be transmitted and randomly grouping it, combined with the demand bandwidth matching model constructed by machine learning, it can more accurately predict and match the actual bandwidth requirements of data slices in long-distance communication, thereby optimizing the selection of transmission links, improving the efficiency and reliability of domestic switches in long-distance communication transmission, and achieving the technical effects of efficient and stable long-distance communication transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 It is a schematic flowchart of a long-distance communication transmission method for a domestic switch provided by the present invention.

[0010] Figure 2 It is a schematic structural diagram of a long-distance communication transmission system for a domestic switch provided by the present invention.

[0011] Description of reference numerals: Initialization module 11, data slicing module 12, model matching module 13, solution acquisition module 14. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0012] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present invention.

[0013] In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the described features. In the description of the present invention, "a plurality" means two or more, unless otherwise specifically defined.

[0014] In the description of the present invention, the term "for example" is used to mean "serving as an example, illustration, or explanation". Any embodiment described as "for example" in the present invention is not necessarily construed as being more preferred or more advantageous than other embodiments. In order for any person skilled in the art to implement and use the present invention, the following description is given. In the following description, details are set forth for the purpose of explanation. It should be understood that those skilled in the art can recognize that the present invention can be implemented without using these specific details. In other instances, well-known structures and processes will not be elaborated in detail to avoid unnecessary details from obscuring the description of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope that conforms to the principles and features disclosed in the present invention.

[0015] Embodiment 1:

[0016] As Figure 1 shown, the embodiment of the present invention provides a long - distance communication transmission method for a domestic switch, which is applied to a domestic switch and includes:

[0017] S10: Obtain the packet size, constraint duration, and target receiving end of the data to be transmitted, and simultaneously initialize the available bandwidths of multiple transmission links with the target receiving end.

[0018] S20: Slice the data to be transmitted to obtain a data slice sequence, where each data slice has the same size, and the data slice has a preset first required bandwidth.

[0019] S30: With the number of transmission links as the maximum group number constraint, randomly group the data slice sequence to obtain several groups of data slices. For each group of data slices, input the number of data slices, the first required bandwidth, and the constraint duration into the required bandwidth matching model, and output several second required bandwidths. The required bandwidth matching model is constructed by machine learning through multiple groups of data. Any one of the multiple groups of data includes: the number of data slice records of a preset communication protocol, the unit slice required record bandwidth, the constraint record duration, and the label indicating the output required bandwidth.

[0020] S40: Calculate the ratios of the several second required bandwidths to the decision ratio threshold respectively to obtain several minimum available bandwidths, perform transmission link matching based on the multiple available bandwidths, and obtain a data transmission scheme to transmit the data to be transmitted to the target receiving end.

[0021] Exemplarily, a domestic switch realizes network communication by receiving, storing, and forwarding data packets. It forwards the data packet to the correct port or link according to the destination address of the data packet. Domestic switches often adopt three - layer switching technology, that is, in addition to having the data link layer forwarding function of a two - layer switch, it also supports the routing forwarding function of the network layer, which enables the switch to forward data packets between different subnets, improving the flexibility and scalability of the network. When the switch is in long - distance communication, due to the large size of the data packets carried, in order to adapt to the bandwidth limitation and latency problems in long - distance transmission, it is necessary to slice the data, cut the large data packet into multiple small data packets for transmission, and then recombine them at the receiving end. This method can reduce the packet loss rate and latency during transmission and improve the data transmission efficiency. However, it is difficult to achieve intelligent bandwidth configuration after data slicing, so a long - distance communication transmission method for a domestic switch is proposed.

[0022] In a specific embodiment, before preparing for data transmission, it is first necessary to obtain the packet size of the data to be transmitted, the constraint duration, and the identifier of the target receiving end. The packet size determines the amount of data transmitted each time, while the constraint duration limits that the data must be successfully transmitted to the target receiving end within this time. The identifier of the target receiving end is used to locate the device receiving the data in the network. At the same time, to ensure the reliability and efficiency of data transmission, multiple transmission links need to be established with the target receiving end. These links can be based on different physical media (such as wired, wireless) or network protocols (such as TCP, UDP). Each link has its specific available bandwidth, that is, the amount of data that can be transmitted per second on this link. Initialize these transmission links and obtain the available bandwidth information of each link, which is usually completed through network measurement or negotiation processes. The role of initialization is to prepare for data transmission, determine the packet size, the constraint duration, and evaluate the available bandwidth of multiple transmission links with the target receiving end. Understanding the bandwidth capabilities of each link is crucial for subsequent data allocation and scheduling. After obtaining the packet size, the constraint duration, the target receiving end, and the available bandwidth information of each transmission link, the data transmission planning phase can be entered. In this phase, it is necessary to formulate a transmission strategy based on the bandwidth capabilities of the links, the priority of the packets, and the constraint duration. During the data transmission process, it is also necessary to continuously monitor the performance and status of each link. If it is found that the performance of a certain link deteriorates or a fault occurs, the transmission strategy can be adjusted in a timely manner, and the data transmission can be switched to other available links to ensure that the data can be successfully transmitted to the target receiving end within the constraint duration. Through the above settings, the efficiency and reliability of data transmission are ensured.

[0023] Furthermore, after determining the overall size of the data packet, the size of each data slice is set according to the network transmission conditions, the bandwidth capacity of the link, and the characteristics of the data packet. This size is fixed, meaning that regardless of the size of the original data packet, it will be cut into a series of data slices of the same size. The purpose of setting the size of each data slice is to optimize the transmission efficiency. Smaller data slices can more flexibly adapt to different transmission links, reducing the risk of transmission delay or packet loss caused by overly large data packets. At the same time, each data slice is assigned a preset first required bandwidth, which is calculated based on factors such as slice size, transmission protocol, and link conditions. The first required bandwidth represents the minimum bandwidth required for the data slice to be transmitted under ideal conditions. Through data slice processing, a data slice sequence is obtained. Each slice in this sequence is a part of the original data packet, and they are processed independently to more effectively utilize the bandwidth resources of the transmission link. The generation of the data slice sequence provides a basis for subsequent operations such as data grouping, bandwidth matching, and link selection. In summary, the process of data slicing is to cut the data to be transmitted into a series of data slices of the same size with a preset first required bandwidth, in order to more efficiently utilize the bandwidth resources of the transmission link and ensure that the data can reach the target receiving end stably and reliably within a given constrained time duration.

[0024] Next, the number of transmission links is used as the maximum group number constraint to randomly group the data slice sequence to be transmitted. This means that if there are N transmission links, the data slice sequence will be randomly divided into N groups (or fewer if the total number of data slices is not enough to fill all the links). The purpose of random grouping is to reduce the problem of low link utilization caused by uneven distribution of data slices and increase the flexibility of the transmission scheme. Then, for each group of data slices, the number of data slices in it, the preset first required bandwidth of each slice, and the constraint duration of the overall transmission are used as inputs and fed into the pre-constructed required bandwidth matching model. The role of this model is to predict and match the actual bandwidth required by each group of data slices during actual transmission, that is, the second required bandwidth. The required bandwidth matching model is a prediction model based on machine learning. Its input features include the number of data slice records under the preset communication protocol, the required record bandwidth of each unit slice, and the constraint duration of the overall transmission. These features jointly describe the transmission requirements and constraint conditions of the data slice group. The output of the model is one or more second required bandwidth values, which represent the actual bandwidth required by each group of data slices under different transmission conditions. The structure of the model includes an input layer, a hidden layer, and an output layer. The input layer receives the above feature data; the hidden layer performs complex non-linear transformations on the data through a multi-layer neural network structure to capture the complex relationships between the features; the output layer outputs the predicted second required bandwidth value. The training of the model is carried out through historical data. Specifically, multiple datasets containing the actual number of transmitted data slices, the required bandwidth of unit slices, the constraint duration, and the labels of the actual required bandwidth for transmission are used to train the model. By continuously adjusting the model parameters, the error between the predicted bandwidth and the actual bandwidth is minimized, thereby improving the prediction accuracy of the model. The evaluation of the model's effect is mainly carried out by comparing the predicted bandwidth with the bandwidth requirements during actual transmission. A high-quality model can accurately predict the bandwidth requirements under different transmission conditions, so as to make more reasonable link selection and bandwidth allocation decisions in the transmission planning stage. In summary, the strategy combining random grouping and the required bandwidth matching model can reasonably allocate data slices according to the number and characteristics of transmission links, predict the actual bandwidth required by each group of data slices, and then a more efficient and reliable transmission scheme can be formulated to ensure that the data can stably reach the target receiver within the given constraint duration.

[0025] Specifically, after obtaining the second required bandwidth, an efficient and feasible data transmission plan needs to be formulated based on these predicted bandwidths and the actual available bandwidth of the network. First, a decision ratio threshold is set, which is used to evaluate the matching degree between the predicted bandwidth and the actual available bandwidth. The decision ratio threshold is usually determined comprehensively according to factors such as network conditions, transmission requirements, and historical experience, and it represents an acceptable bandwidth utilization ratio. Then, the ratio of the second required bandwidth of each group of data slices to the decision ratio threshold is calculated respectively. This ratio reflects the occupancy of the transmission link bandwidth resources by each group of data slices on the premise of meeting the transmission requirements. By comparing these ratios, it is possible to initially screen out which groups of data slices are more likely to obtain sufficient bandwidth resources under the current network conditions, so as to achieve efficient transmission. Next, the transmission links are matched based on the available bandwidth information of multiple transmission links. By selecting the most suitable transmission link according to the bandwidth requirements of the data slice groups, links with sufficient bandwidth, low latency, and high stability can be given priority to ensure that the data can reach the target receiving end stably within the given constraint duration. At the same time, the load balancing problem of the links also needs to be considered during the link matching process. In order to avoid performance degradation of a certain link due to carrying too many data slices, polling, weighted polling, or other load balancing algorithms can be used to allocate the data slice groups to different links. A detailed data transmission plan is formulated according to the results of the link matching. This plan includes information such as which link each group of data slices should be transmitted through, the transmission order, and the bandwidth allocation of each link. By executing this plan, the data to be transmitted can be efficiently and reliably transmitted to the target receiving end. In summary, by calculating the ratio of the second required bandwidth to the decision ratio threshold and the step of matching the transmission links based on the available bandwidth, an efficient and feasible data transmission plan can be formulated to ensure that the data can reach the target receiving end stably within the given constraint duration.

[0026] This solution realizes the refined management and optimized allocation of the data to be transmitted through key links such as data slicing, random grouping, required bandwidth prediction, bandwidth ratio calculation, and transmission link matching. This process significantly improves the utilization rate of network resources and ensures that the data can be stably and efficiently transmitted to the target receiving end within the given constraint duration. At the same time, by introducing a machine learning model for required bandwidth prediction, the flexibility and adaptability of the transmission plan are enhanced, providing strong technical support for data transmission in complex network environments.

[0027] In a preferred embodiment, initializing the available bandwidths of multiple transmission links with a target receiving end includes: receiving the load states of multiple transmission links periodically fed back by the target receiving end; obtaining the multiple communication protocol types and multiple bandwidth occupancy rates of the multiple transmission links; extracting the first load state, the first communication protocol type, and the first bandwidth occupancy rate of a first transmission link from the multiple load state information, the multiple communication protocol types, and the multiple bandwidth occupancy rates; retrieving the mode available bandwidth of a sample transmission channel that simultaneously has the first load state, the first communication protocol type, and the first bandwidth occupancy rate, setting it as the first available bandwidth, and adding it to the multiple available bandwidths.

[0028] Optionally, in the initialization stage of data transmission, in order to accurately evaluate and initialize the available bandwidth of multiple transmission links connected to the target receiver, a set of methods based on big data analysis are adopted. This method aims to comprehensively consider multiple dimensions such as the load status of the link, the type of communication protocol, and the bandwidth occupancy rate, so as to more accurately determine the actual available bandwidth of each link and provide a solid data foundation for subsequent data transmission link scheduling. Specifically, the receiving end receives the load status information of multiple transmission links periodically feedback by the target receiver, and these information reflect the real-time usage of the link, including but not limited to the busy degree of the link, the data transmission rate, etc. By continuously collecting these load status information, the operating condition of the link can be grasped in real time. Further obtain the communication protocol type and bandwidth occupancy rate of multiple transmission links. The communication protocol type determines the way and efficiency of data transmission, while the bandwidth occupancy rate directly reflects the proportion of the bandwidth resources currently occupied by the link. Furthermore, the load status, communication protocol type, and bandwidth occupancy rate of a specific transmission link (i.e., the first transmission link) are extracted from the collected multiple load status information, communication protocol type, and bandwidth occupancy rate. This step is to conduct a detailed analysis and evaluation for a single link. After extracting the relevant information of the first transmission link, use big data analysis technology to retrieve samples in the existing sample transmission channel database that have the same load status, communication protocol type, and bandwidth occupancy rate as the first transmission link. These sample transmission channels represent a set of links with similar characteristics to the first transmission link. Then, extract the mode (i.e., the value with the highest frequency of occurrence) of the available bandwidth from the retrieved sample transmission channels and set it as the first available bandwidth of the first transmission link. This step is based on the statistical law of big data, believing that the mode available bandwidth can better reflect the actual available bandwidth situation of this type of link. Subsequently, add this first available bandwidth to the set of multiple available bandwidths for subsequent use in link scheduling. Due to the large number and complexity of input variable elements, directly using model fitting may face problems such as convergence difficulties. Therefore, by choosing a method based on big data analysis to retrieve the theoretical available bandwidth. This method not only considers multiple dimensional information of the link but also makes full use of the statistical law of big data, thus being able to more accurately evaluate the available bandwidth of the link and providing strong data support for subsequent data transmission link scheduling.

[0029] In a preferred embodiment, retrieving the mode available bandwidth of sample transmission channels that simultaneously have the first load state, the first communication protocol type, and the first bandwidth occupancy rate includes: the load state includes throughput, latency, latency variance, and packet loss rate; configuring a first-level fault tolerance neighborhood for throughput, latency, latency variance, packet loss rate, and the first bandwidth occupancy rate; retrieving a set of first-level available bandwidth record values of first-level sample transmission channels that simultaneously have the first communication protocol type and meet the first-level fault tolerance neighborhood, where any first-level available bandwidth record value in the set of first-level available bandwidth record values has a first-level load state identifier and a first-level bandwidth occupancy rate identifier; configuring a second-level fault tolerance neighborhood according to the first-level load state identifier and the first-level bandwidth occupancy rate identifier; retrieving a set of second-level available bandwidth record values of second-level sample transmission channels that simultaneously have the first communication protocol type and meet the second-level fault tolerance neighborhood; performing a mode evaluation on the set of first-level available bandwidth record values and the set of second-level available bandwidth record values to obtain the mode available bandwidth.

[0030] Furthermore, during the evaluation of the available bandwidth of the data transmission link, in order to ensure that sufficiently accurate and reliable bandwidth data can be obtained, a method of hierarchical retrieval and fault tolerance processing is adopted. This method aims to improve the accuracy of available bandwidth evaluation through meticulous data screening and fault tolerance mechanisms. First, clarify four key indicators of the load status: throughput, latency, latency variance, and packet loss rate. Among them, throughput, also known as the overall packet forwarding rate, refers to the quantity of data successfully transmitted by a network, device, port, or other facilities within a unit of time, usually measured in bits (bit) or bytes (byte). In the absence of frame loss, throughput represents the maximum data rate that a device can receive and forward. It is an extreme indicator that reflects the performance of network devices when all ports are fully configured and operating at the highest line speed. The magnitude of throughput is mainly determined by the hardware of the internal and external network interfaces of the network device and the efficiency of the program algorithm. Latency refers to the time elapsed for a packet to travel from one device to another. During data transmission, latency may include multiple components such as transmission latency, propagation latency, and processing latency. Transmission latency refers to the time required for data to start transmitting a data frame to the completion of data frame transmission at the sending end, which is related to the length of the data frame and the sending rate. Propagation latency refers to the time required for the sending end to start sending data to the receiving end to receive the data, which is related to the transmission distance. Latency is an important indicator for measuring network response speed. Low latency means faster data transmission and response. Latency variance is an indicator for measuring the degree of latency fluctuation. In an actual network, due to the influence of various factors (such as network congestion, device failures, etc.), latency may change. Latency variance reflects the degree of this change, that is, the difference in the delay time of adjacent data packets. Low latency variance means more stable network performance and smoother data transmission. Packet loss rate refers to the proportion of the number of lost data packets in the total number of transmitted data packets during data transmission. When data packets are transmitted in the network, they may be lost due to various reasons (such as network congestion, device failures, signal interference, etc.). The level of packet loss rate directly affects the reliability and integrity of data transmission. A high packet loss rate will lead to problems such as data retransmission and transmission delay, seriously affecting network performance. The above indicators together constitute a comprehensive description of the link performance. Then, in order to handle the possible fluctuations and errors in the actual data, a first-level fault tolerance neighborhood is configured for throughput, latency, latency variance, packet loss rate, and the first bandwidth occupancy rate. These fault tolerance neighborhoods define the acceptable range of data fluctuations, ensuring that the actual changes in the data can be taken into account when retrieving samples. The first-level fault tolerance neighborhood can be set according to the actual situation. It should be noted that when setting, the application scenario and the mutual influence between different performance indicators should be comprehensively considered. After that, the retrieval of the first-level sample transmission channel begins. Among the samples with the first communication protocol type, those samples whose load status and bandwidth occupancy rate both fall within the first-level fault tolerance neighborhood are screened out to form a set of first-level sample transmission channels.For each sample in this set, record its primary available bandwidth, the corresponding primary load status identifier, and the primary bandwidth occupancy rate identifier. To improve the accuracy and robustness of the available bandwidth assessment, a secondary fault-tolerant neighborhood is further configured. This fault-tolerant neighborhood appropriately relaxes the fluctuation ranges of the load status and bandwidth occupancy rate based on the primary fault-tolerant neighborhood. By relaxing the fault-tolerant range, it is expected to retrieve more sample data, thereby increasing the data volume and diversity. Subsequently, a search for the secondary sample transmission channels is conducted. Among the samples with the first communication protocol type, those samples whose load status and bandwidth occupancy rate fall within the secondary fault-tolerant neighborhood are selected to form a set of secondary sample transmission channels. Each sample in this set records its secondary available bandwidth. Finally, a mode evaluation is performed on the sets of available bandwidth recorded values of the primary and secondary sample transmission channels. Mode evaluation is a method based on statistical laws, which believes that the value with the highest occurrence frequency (i.e., the mode) can better reflect the true situation of the data. By performing mode evaluation on the available bandwidth recorded values in the primary and secondary sets, the final mode available bandwidth can be obtained. This mode available bandwidth not only considers the exact matching of the data (primary fault-tolerant neighborhood) but also takes into account the diversity and robustness of the data (secondary fault-tolerant neighborhood), thereby improving the accuracy of the available bandwidth assessment.

[0031] In a preferred embodiment, the required bandwidth matching model is constructed through machine learning with multiple sets of data. Any one of the multiple sets of data includes: the number of data slice records of a preset communication protocol, the required bandwidth recorded per unit slice, the constrained recording duration, and a label identifying the required bandwidth. Specifically, with the label identifying the required bandwidth as the supervision, and using the number of data slice records, the required bandwidth recorded per unit slice, and the constrained recording duration as inputs, a neural network is trained to generate a primary required bandwidth matching model; the primary loss data with an output accuracy rate of the primary required bandwidth matching model less than or equal to the output accuracy rate threshold is extracted; when the primary loss data is greater than or equal to the convergence loss amount threshold, the loss calculation weight of the primary loss data in the multiple sets of data is increased, and the neural network is trained to obtain a secondary required bandwidth matching model; until the N-level loss data is less than the convergence loss amount threshold, the mean values of the outputs of the primary, secondary, up to the N-level required bandwidth matching models are deployed to the output layer to generate the primary required bandwidth matching model.

[0032] Specifically, in the process of constructing the demand bandwidth matching model, an iterative and hierarchical improvement method is adopted to optimize the prediction performance of the model. First, multiple groups of data are collected. Each group of data contains four key elements: the number of data slice records of the preset communication protocol, the recorded bandwidth of the unit slice demand, the constrained recording duration, and a label indicating the output demand bandwidth. These labels serve as supervision information to guide the training process of the model. Using this prepared data, with the label indicating the output demand bandwidth as the supervision target, and the number of data slice records, the recorded bandwidth of the unit slice demand, and the constrained recording duration as input features, a neural network is trained to generate a first-level demand bandwidth matching model. This model initially has the ability to predict the demand bandwidth based on the input features. Next, the output accuracy of the first-level model is evaluated, and the data with an output accuracy less than or equal to the preset output accuracy threshold is identified. This data is called the first-level loss data. These loss data reflect the deficiencies in the model's prediction. If the number of first-level loss data is greater than or equal to the convergence loss quantity threshold, it means that there is still a large error space in the model's prediction. To optimize this part of the error, the loss calculation weight of these first-level loss data in multiple groups of data is increased, so that the model pays more attention to these difficult-to-predict data points in subsequent training. Subsequently, the neural network is retrained based on the adjusted weights to obtain a second-level demand bandwidth matching model. The second-level model aims to fit the part that the first-level model fails to accurately predict, that is, the error data of the first-level model. Repeat the above steps, continuously extract higher-level loss data, adjust the weights, and train higher-level demand bandwidth matching models until the N-level loss data is less than the convergence loss quantity threshold. This process is an iterative optimization process. Each level of the model attempts to fit the error of the previous level of the model, thereby gradually improving the prediction accuracy of the overall model. Finally, the outputs of the first-level, second-level, and up to N-level demand bandwidth matching models are averaged and deployed in the output layer to generate the final demand bandwidth matching model. This integration method utilizes the prediction results of multiple-level models and reduces the bias and variance of a single model through averaging, improving the generalization ability and prediction accuracy of the overall model. In summary, the entire process realizes the gradual improvement of the model performance through the construction of multi-level models and by using the next-level model to fit the error data of the previous level. Finally, by integrating the outputs of multi-level models, a more accurate and stable demand bandwidth matching model is generated. This method not only improves the prediction accuracy of the model but also enhances the model's adaptability to complex data.

[0033] In a preferred embodiment, the ratios of the several second required bandwidths to the decision ratio threshold are calculated respectively to obtain several minimum available bandwidths, and transmission link matching is performed based on the multiple available bandwidths to obtain a data transmission scheme, including: randomly configuring several transmission links from the multiple available bandwidths, where each available bandwidth is greater than or equal to the several minimum available bandwidths, to construct a first transmission link combination; until all transmission link combinations that meet the several available bandwidths among the multiple available bandwidths are enumerated, to obtain the Mth transmission link combination; performing sorting with the maximum matching degree of transmission order on the first transmission link combination and the Mth transmission link combination to obtain the data transmission scheme.

[0034] Optionally, several bandwidths are randomly selected from multiple available bandwidths to ensure that these bandwidths are respectively greater than or equal to the several minimum available bandwidths calculated previously. In this way, each required bandwidth can find at least one link that meets its transmission requirements. These links are combined to construct a first transmission link combination. To find the optimal transmission scheme, all possible transmission link combinations need to be enumerated. This is achieved by continuously repeating the above random configuration process until all possible transmission link combinations are considered, to obtain the Mth transmission link combination (M represents the total number of all possible combinations). After obtaining all possible transmission link combinations, these combinations need to be evaluated to determine the optimal transmission scheme. Here, the method of sorting with the maximum matching degree of transmission order is adopted, that is, considering the transmission order of data and the bandwidth size of the link. Since a link with a larger bandwidth can transmit more data within the same time period, the data sorted earlier (which may be determined according to the importance, urgency, or other factors of the data) is preferentially deployed on the link with a larger bandwidth. Through this method, it can be ensured that critical data is preferentially transmitted, and at the same time, the transmission capacity of the link is fully utilized. Finally, the optimal data transmission scheme is determined according to the result of sorting with the maximum matching degree of transmission order. This scheme not only meets the requirements of all required bandwidths, but also optimizes the transmission order of data and the utilization of link bandwidth, thereby improving the overall data transmission efficiency and performance. In summary, through steps such as calculating the minimum available bandwidth, randomly configuring transmission links, enumerating all possible transmission link combinations, and sorting with the maximum matching degree of transmission order, the transmission link matching based on multiple available bandwidths and the formulation of the data transmission scheme are realized. This method fully considers the influence of bandwidth size and data transmission order on transmission efficiency. The disadvantage of the traditional multi-line transmission scheme is that the data may arrive out of order, resulting in a relatively large buffer pressure. Through the above optimization method, considering the allocation of transmitted data according to the speed of transmission efficiency can reduce the probability of out-of-order arrival, thereby reducing the buffer pressure.

[0035] In a preferred embodiment, the first transmission link combination and the Mth transmission link combination are sorted by the maximum matching degree of transmission sequence to obtain the data transmission scheme, including: sorting the plurality of groups of data slices according to the data slice sequence to obtain a plurality of groups of data slice sorting results; sorting the available bandwidth of the first transmission link combination in descending order to obtain a first sorting result, adjusting the sequence number position of the first sorting result based on the first transmission link combination to obtain a first sequence number sequence; adjusting the sequence number position of the plurality of groups of data slice sorting results based on the first transmission link combination to obtain a first data slice sequence number sequence; sorting the available bandwidth of the Mth transmission link combination in descending order to obtain Obtain an Mth sorting result, adjust the sequence number position of the Mth sorting result based on the Mth transmission link combination, and obtain an Mth sequence number sequence; adjust the sequence number positions of several groups of data slice sorting results based on the Mth transmission link combination, and obtain an Mth data slice sequence number sequence; compare the first sequence number sequence with the sequence numbers of the first data slice sequence, count the proportion of consistent sequence numbers, and set it as the first combination transmission order matching degree; until the sequence numbers of the Mth sequence number sequence with the sequence numbers of the Mth data slice sequence are compared, count the proportion of consistent sequence numbers, and set it as the Mth combination transmission order matching degree; extract the first combination transmission order matching degree until the transmission link combination corresponding to the maximum value of the Mth combination transmission order matching degree is set as the data transmission scheme.

[0036] Further, several groups of data slices are sorted according to a predetermined data slice sequence, which is usually based on the importance, urgency or other business logics of the data. After sorting, the sorting results of several groups of data slices are obtained, and these results reflect the priority transmission order of the data slices. For each transmission link combination (from the first transmission link combination to the Mth transmission link combination), the available bandwidths it contains are sorted in descending order. This step is to identify the link with the largest bandwidth in each combination, because a link with a larger bandwidth can transmit data faster. Based on the sorted transmission link combinations, the serial number positions of their available bandwidths are adjusted to form a serial number sequence. At the same time, based on the sorted data slice results, the serial number positions of the data slices are adjusted to form a data slice serial number sequence. These serial number sequences are used for subsequent comparison and analysis. Next, for each transmission link combination, its serial number sequence is compared with the corresponding data slice serial number sequence. Specifically, it is checked whether the serial numbers of the same positions in the two sequences are the same. If they are the same, it means that the link bandwidth at this position matches the priority transmission order of the data slice. The proportion of all the same serial numbers is counted, and this proportion is the transmission order matching degree. The above comparison and counting processes need to be repeated until all transmission link combinations (from the first combination to the Mth combination) have completed the comparison with the data slice serial number sequence. In this way, the transmission order matching degree of each transmission link combination can be obtained. Finally, the maximum value is extracted from the transmission order matching degrees of all transmission link combinations. The transmission link combination corresponding to this maximum value is considered the optimal data transmission scheme. Because it achieves the best match in bandwidth allocation and data slice transmission order, it can most effectively utilize the transmission resources and improve the efficiency and performance of data transmission. In summary, through steps such as data slice sorting, transmission link combination sorting, serial number adjustment, comparison and statistics, and optimal scheme determination, the matching degree analysis of the transmission link combination and the data slice sorting results is realized. This method can comprehensively consider the influence of bandwidth size and data transmission order, so as to formulate a more efficient and optimized data transmission scheme.

[0037] In a preferred embodiment, when any one of several minimum available bandwidths is greater than the maximum available bandwidth of the multiple available bandwidths, with the number of transmission links as the maximum group number constraint, the data slice sequence is randomly grouped again to obtain several groups of updated data slices; a loop is executed based on the several groups of updated data slices.

[0038] Specifically, when faced with a situation where any one of several minimum available bandwidths exceeds the maximum available bandwidth among multiple existing available bandwidths, it means that the current bandwidth resources cannot meet the minimum requirements for data transmission. To solve this problem and find a feasible transmission scheme, it is necessary to reconfigure the data slices and transmission links. First, compare several minimum available bandwidths with the maximum value of multiple available bandwidths. This is to determine whether the current bandwidth resources are sufficient to support the minimum requirements for data transmission. If it is found that any one of the minimum available bandwidths is greater than the maximum available bandwidth, it means that the current combination of transmission links is invalid. Because no single link can provide sufficient data transmission capacity to meet the requirements. After identifying the invalid combination, it is necessary to randomly group the data slice sequence again based on the maximum number of groups constraint of the transmission links. The purpose of doing this is to try different data slice configurations in order to find a transmission scheme that can adapt to the current bandwidth resources. Through random grouping, several updated data slice groups are obtained. These updated data slice groups will be used in the subsequent transmission link matching and evaluation processes. Subsequently, based on the updated data slice groups, the same process as before is executed, including calculating the minimum available bandwidth, constructing a combination of transmission links, enumerating all possible combinations, sorting the combinations according to the maximum matching degree of the transmission order, and so on. During the loop execution process, continuously evaluate the matching degree of the transmission order of each transmission link combination, and select the optimal transmission scheme according to the result of the matching degree. If a scheme that meets the requirements is found, the process ends; if not, continue to execute the loop until a feasible scheme is found or the preset loop count limit is reached. The above method reflects the flexibility and adaptability of scheme optimization, and can find an effective transmission strategy under the condition of limited bandwidth resources.

[0039] The long-distance communication transmission method of a domestic switch provided by the embodiment of the present invention has at least the following technical effects:

[0040] 1. By introducing a demand bandwidth matching model, it is possible to intelligently output a second demand bandwidth that adapts to different transmission requirements based on parameters such as the number of data slices, the demand bandwidth per unit slice, and the constraint duration. This intelligent bandwidth matching mechanism not only improves the utilization efficiency of bandwidth resources, but also ensures the stability and efficiency of data transmission. At the same time, by sorting multiple transmission links according to the maximum matching degree of the transmission order, the data transmission scheme is further optimized, making the data transmission smoother and faster.

[0041] 2. By initializing the available bandwidths of multiple transmission links with the target receiver and receiving information such as the load status periodically feedback by the target receiver in real time, dynamic management and optimized allocation of bandwidth resources are achieved. When it is detected that the minimum available bandwidth exceeds the maximum value of the current available bandwidth, the data slice sequence can be automatically randomly grouped again, and a loop is executed based on the updated data slices to adapt to different transmission environments and requirements. This dynamic management and enhanced adaptability can flexibly handle various complex transmission scenarios, improving the reliability and stability of data transmission.

[0042] 3. By slicing the data to be transmitted and randomly grouping based on the maximum number of groups constraint of the number of transmission links, an effective combination of data slices and transmission links is achieved. At the same time, by comparing the transmission order matching degrees of different transmission link combinations, the optimal data transmission scheme is selected. The impacts of bandwidth size and data transmission order on transmission efficiency are fully considered. The disadvantage of traditional multi-line transmission schemes is that they may arrive out of order, resulting in a large buffer pressure. Through the above optimization method, considering the allocation of transmitted data according to the transmission efficiency, the probability of out-of-order arrival can be reduced, thereby reducing the buffer pressure.

[0043] Embodiment 2:

[0044] As Figure 2 shown, based on the same inventive concept as the long-distance communication transmission method of a domestic switch provided in Embodiment 1, the present invention embodiment also provides a long-distance communication transmission system for a domestic switch, including:

[0045] An initialization module 11, configured to obtain the packet size, constraint duration, and target receiver of the data to be transmitted, and at the same time initialize the available bandwidths of multiple transmission links with the target receiver.

[0046] A data slicing module 12, configured to slice the data to be transmitted to obtain a data slice sequence, where each data slice has the same size, and the data slice has a preset first required bandwidth.

[0047] A model matching module 13, configured to randomly group the data slice sequence with the number of transmission links as the maximum number of groups constraint to obtain several groups of data slices. For each group of data slices, the number of data slices, the first required bandwidth, and the constraint duration are respectively input into a required bandwidth matching model, and several second required bandwidths are output. The required bandwidth matching model is constructed by machine learning through multiple groups of data, and any one of the multiple groups of data includes: the number of data slice records of a preset communication protocol, the bandwidth of the unit slice requirement record, the constraint record duration, and a label indicating the output required bandwidth.

[0048] A solution acquisition module 14 is configured to calculate the ratios of the several second required bandwidths to the decision ratio threshold respectively to obtain several minimum available bandwidths, perform transmission link matching based on the multiple available bandwidths, obtain a data transmission solution, and transmit the data to be transmitted to a target receiver.

[0049] Furthermore, the initialization module 11 is further configured to perform the following steps:

[0050] Obtain the packet size, constraint duration, and target receiver of the data to be transmitted, and initialize the multiple available bandwidths of the multiple transmission links to the target receiver; slice the data to be transmitted to obtain a data slice sequence, where each data slice has the same size, and the data slice has a preset first required bandwidth; perform random grouping on the data slice sequence with the number of transmission links as the maximum group number constraint to obtain several groups of data slices, and for each group of data slices, input the number of data slices, the first required bandwidth, and the constraint duration into a required bandwidth matching model respectively to output several second required bandwidths. The required bandwidth matching model is constructed by machine learning through multiple groups of data. Any one of the multiple groups of data includes: the number of data slice records of a preset communication protocol, the unit slice required recorded bandwidth, the constraint recorded duration, and a label indicating the output required bandwidth; calculate the ratios of the several second required bandwidths to the decision ratio threshold respectively to obtain several minimum available bandwidths, perform transmission link matching based on the multiple available bandwidths, obtain a data transmission solution, and transmit the data to be transmitted to a target receiver.

[0051] Furthermore, the initialization module 11 is further configured to perform the following steps:

[0052] The load status includes throughput, delay, delay variance, and packet loss rate; configure a first-level fault tolerance neighborhood for throughput, delay, delay variance, packet loss rate, and the first bandwidth occupancy rate; retrieve the set of first-level available bandwidth record values of the first-level sample transmission channels that have the first communication protocol type and meet the first-level fault tolerance neighborhood. Any one of the first-level available bandwidth record values in the set of first-level available bandwidth record values has a first-level load status identifier and a first-level bandwidth occupancy rate identifier; configure a second-level fault tolerance neighborhood according to the first-level load status identifier and the first-level bandwidth occupancy rate identifier; retrieve the set of second-level available bandwidth record values of the second-level sample transmission channels that have the first communication protocol type and meet the second-level fault tolerance neighborhood; perform a mode evaluation on the set of first-level available bandwidth record values and the set of second-level available bandwidth record values to obtain the mode available bandwidth.

[0053] Furthermore, the model matching module 13 is further configured to perform the following steps:

[0054] The label identifying the output required bandwidth is used as supervision, and the number of data slice records, the required record bandwidth per unit slice, and the constraint record duration are used as input to train a neural network to generate a first-level required bandwidth matching model; the first-level loss data whose output accuracy of the first-level required bandwidth matching model is less than or equal to the output accuracy threshold is extracted; when the first-level loss data is greater than or equal to the convergence loss threshold, the loss calculation weight of the first-level loss data in the multiple groups of data is gained, and the neural network is trained to obtain a second-level required bandwidth matching model; until the N-level loss data is less than the convergence loss threshold, the output mean of the first-level, second-level, and N-level required bandwidth matching models is deployed in the output layer to generate the first-level required bandwidth matching model.

[0055] Furthermore, the solution acquisition module 14 is further configured to perform the following steps:

[0056] A plurality of transmission links whose available bandwidths are respectively greater than or equal to the plurality of minimum available bandwidths are randomly configured from the plurality of available bandwidths to construct a first transmission link combination; until the plurality of transmission link combinations whose available bandwidths satisfy the plurality of available bandwidths are completely enumerated, an Mth transmission link combination is obtained; and the first transmission link combination and the Mth transmission link combination are sorted by the maximum matching degree of transmission sequence to obtain the data transmission scheme.

[0057] Furthermore, the solution acquisition module 14 is further configured to perform the following steps:

[0058] The plurality of groups of data slices are sorted according to the data slice sequence to obtain sorting results of the plurality of groups of data slices; the available bandwidths of the first transmission link combination are sorted in descending order to obtain a first sorting result, and the sequence number position of the first sorting result is adjusted based on the first transmission link combination to obtain a first sequence number sequence; the sequence number positions of the plurality of groups of data slice sorting results are adjusted based on the first transmission link combination to obtain a first sequence number sequence; the available bandwidths of the Mth transmission link combination are sorted in descending order to obtain an Mth sorting result, and the sequence number positions of the Mth sorting result are adjusted based on the Mth transmission link combination. The sequence number position of the data slice sorting results is adjusted based on the Mth transmission link combination to obtain the Mth data slice sequence number sequence; the sequence numbers of the first sequence number sequence and the first data slice sequence number sequence are compared, and the proportion of the consistent sequence numbers is counted, which is set as the first combination transmission order matching degree; until the sequence numbers of the Mth sequence number sequence and the Mth data slice sequence number sequence are compared, the proportion of the consistent sequence numbers is counted, which is set as the Mth combination transmission order matching degree; the transmission order matching degree of the first combination is extracted until the transmission link combination corresponding to the maximum value of the transmission order matching degree of the Mth combination is set as the data transmission scheme.

[0059] Furthermore, the solution acquisition module 14 is further configured to perform the following steps:

[0060] When any one of several minimum available bandwidths is greater than the maximum available bandwidth of the multiple available bandwidths, with the number of transmission links as the maximum group number constraint, the data slice sequence is randomly grouped again to obtain several groups of updated data slices; a loop is executed based on the several groups of updated data slices.

[0061] Through the foregoing detailed description of a long-distance communication transmission method for a domestic switch in this specification, those skilled in the art can clearly know a long-distance communication transmission system for a domestic switch in this embodiment. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple. For related parts, refer to the description in the method section.

[0062] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A long-distance communication transmission method for a domestic switch, characterized in that, Applied to domestic switches, including: Obtaining a data packet size, a constraint duration, and a target receiving end of the data to be transmitted, and simultaneously initializing multiple available bandwidths of multiple transmission links with the target receiving end; Slicing the data to be transmitted to obtain a data slice sequence, each data slice having the same size, wherein the data slice has a preset first required bandwidth; Taking the number of transmission links as the maximum group number constraint, the data slice sequence is randomly grouped to obtain several groups of data slices, and for each group of data slices, the number of data slices, the first required bandwidth and the constraint duration are respectively input into the required bandwidth matching model, and several second required bandwidths are output, wherein the required bandwidth matching model is constructed by machine learning through multiple groups of data, and any group of the multiple groups of data includes: the number of data slice records of a preset communication protocol, the required record bandwidth per unit slice, the constraint record duration and a label identifying the output required bandwidth; Calculating the ratios of the plurality of second required bandwidths to the decision ratio thresholds respectively to obtain a plurality of minimum available bandwidths, performing transmission link matching based on the plurality of available bandwidths, and obtaining a data transmission scheme to transmit the data to be transmitted to a target receiving end; The method comprises: calculating the ratios of the plurality of second required bandwidths to the decision ratio thresholds respectively, obtaining a plurality of minimum available bandwidths, performing transmission link matching based on the plurality of available bandwidths, and obtaining a data transmission scheme, including: Randomly configuring a plurality of transmission links whose available bandwidths are respectively greater than or equal to the plurality of minimum available bandwidths from the plurality of available bandwidths to construct a first transmission link combination; Until the transmission link combinations whose multiple available bandwidths satisfy the plurality of available bandwidths are completely enumerated, an Mth transmission link combination is obtained; performing transmission sequence maximum matching degree sorting on the first transmission link combination to the Mth transmission link combination to obtain the data transmission scheme; The step of performing maximum matching degree sorting of the transmission order of the first transmission link combination to the Mth transmission link combination to obtain the data transmission scheme includes: Sorting the plurality of groups of data slices according to the data slice sequence to obtain sorting results of the plurality of groups of data slices; Sorting the available bandwidths of the first transmission link combinations in descending order to obtain a first sorting result, and adjusting the sequence number position of the first sorting result based on the first transmission link combination to obtain a first sequence number sequence; Adjusting the sequence number positions of the sorting results of the plurality of groups of data slices based on the first transmission link combination to obtain a first data slice sequence number sequence; Sorting the available bandwidths of the Mth transmission link combination in descending order to obtain an Mth sorting result, and adjusting the sequence number position of the Mth sorting result based on the Mth transmission link combination to obtain an Mth sequence number sequence; Adjusting the sequence number positions of the sorting results of the plurality of groups of data slices based on the Mth transmission link combination to obtain the Mth data slice sequence number sequence; Compare the same-digit sequence numbers of the first sequence number sequence and the first data slice sequence number sequence, count the proportion of the same sequence numbers, and set it as the first combined transmission order matching degree; Until comparing the same - digit serial numbers of the M - th serial number sequence and the M - th data slice serial number sequence, count the proportion of consistent serial numbers, which is set as the M - th combined transmission order matching degree; Extract the transmission link combination corresponding to the maximum value from the first combined transmission order matching degree to the M - th combined transmission order matching degree and set it as the data transmission scheme.

2. The method according to claim 1, characterized in that, Initialize the available bandwidths of multiple transmission links with the target receiving end, including: Receive the load status information of multiple transmission links periodically fed back by the target receiving end; Obtain the communication protocol types and bandwidth occupancy rates of multiple transmission links; Extract the first load status, the first communication protocol type, and the first bandwidth occupancy rate of the first transmission link from the multiple load status information, the multiple communication protocol types, and the multiple bandwidth occupancy rates; Retrieve the mode available bandwidth of the sample transmission channels that simultaneously have the first load status, the first communication protocol type, and the first bandwidth occupancy rate, set it as the first available bandwidth, and add it to the multiple available bandwidths.

3. The method according to claim 2, characterized in that, Retrieve the mode available bandwidth of the sample transmission channels that simultaneously have the first load status, the first communication protocol type, and the first bandwidth occupancy rate, including: The load status includes throughput, delay, delay variance, and packet loss rate; Configure a first - level fault - tolerance neighborhood for throughput, delay, delay variance, packet loss rate, and the first bandwidth occupancy rate; Retrieve the set of first - level available bandwidth record values of the first - level sample transmission channels that simultaneously have the first communication protocol type and meet the first - level fault - tolerance neighborhood, where any first - level available bandwidth record value in the set of first - level available bandwidth record values has a first - level load status identifier and a first - level bandwidth occupancy rate identifier; Configure a second - level fault - tolerance neighborhood according to the first - level load status identifier and the first - level bandwidth occupancy rate identifier; Retrieve the set of second - level available bandwidth record values of the second - level sample transmission channels that simultaneously have the first communication protocol type and meet the second - level fault - tolerance neighborhood; Perform a mode evaluation on the set of first - level available bandwidth record values and the set of second - level available bandwidth record values to obtain the mode available bandwidth.

4. The method according to claim 1, wherein The demand bandwidth matching model is constructed by machine learning through multiple groups of data. Any one of the multiple groups of data includes: the number of data slice records of a preset communication protocol, the required bandwidth per unit slice record, the constrained record duration, and a label indicating the required bandwidth, including: Using the label indicating the required bandwidth as supervision, and using the number of data slice records, the required bandwidth per unit slice record, and the constrained record duration as inputs, train a neural network to generate a first - level demand bandwidth matching model; Extract the first - level loss data whose output accuracy rate of the first - level demand bandwidth matching model is less than or equal to the output accuracy rate threshold; When the first - level loss data is greater than or equal to the convergence loss amount threshold, increase the loss calculation weight of the first - level loss data in the multiple groups of data, train the neural network, and obtain a second - level demand bandwidth matching model; Until the loss data at level N is less than the convergence loss amount threshold, the outputs of the demand bandwidth matching models from level 1 to level N are averaged and deployed to the output layer to generate the final demand bandwidth matching model.

5. The method according to claim 1, characterized in that, It further includes: When any one of several minimum available bandwidths is greater than the maximum available bandwidth of the multiple available bandwidths, with the number of transmission links as the maximum group number constraint, the data slice sequence is randomly grouped again to obtain several groups of updated data slices; Execute a loop based on the several groups of updated data slices.

6. A long-distance communication transmission system for a domestic switch, characterized in that, Applied to a domestic switch, used to implement a long-distance communication transmission method for a domestic switch according to any one of claims 1-5, including: An initialization module, used to obtain the packet size, constraint duration, and target receiving end of the data to be transmitted, and at the same time initialize the multiple available bandwidths of the multiple transmission links with the target receiving end; A data slicing module, used to slice the data to be transmitted to obtain a data slice sequence, where each data slice has the same size, and the data slice has a preset first demand bandwidth; A model matching module, used to randomly group the data slice sequence with the number of transmission links as the maximum group number constraint to obtain several groups of data slices. For each group of data slices, the number of data slices, the first demand bandwidth, and the constraint duration are respectively input into the demand bandwidth matching model, and several second demand bandwidths are output. The demand bandwidth matching model is constructed by machine learning through multiple groups of data, and any one of the multiple groups of data includes: the number of data slice records of a preset communication protocol, the bandwidth required for unit slice records, the constraint record duration, and a label indicating the output demand bandwidth; A solution acquisition module, used to calculate the ratios of the several second demand bandwidths to the decision ratio threshold respectively to obtain several minimum available bandwidths, perform transmission link matching based on the multiple available bandwidths, and obtain a data transmission solution to transmit the data to be transmitted to the target receiving end.

Images