Network congestion prediction method and device, electronic equipment and storage medium

Abstract

The application provides a network congestion prediction method and device, electronic equipment and storage medium, and relates to the technical field of quantum communication. The method comprises the following steps: acquiring real-time network state data of multiple dimensions of a quantum cryptography network; predicting the real-time network state data of each dimension by using a hybrid prediction model of each dimension to obtain first network state data and second network state data of each dimension at a prediction time, wherein the first network state data and the second network state data are prediction results of a first prediction model and a second prediction model in the hybrid prediction model; calculating target network state data of each dimension at the prediction time according to the first network state data and the second network state data of each dimension at the prediction time; and determining a congestion probability of the quantum cryptography network at the prediction time according to the target network state data of the multiple dimensions. The application can predict the network congestion of the quantum cryptography network and improve the performance and reliability of the quantum cryptography network.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of quantum communication, in particular to a network congestion prediction method and device, electronic equipment and storage medium. BACKGROUND

[0002] Quantum Key Distribution (QKD) technology is a secure key distribution technology relying on the basic principles of quantum mechanics, which is used to ensure the high security of communication and has great application prospects in the field of information security.

[0003] However, with the popularization of QKD technology in practical application, the congestion problem in QKD network gradually emerges. The traditional network congestion problem also exists in QKD network. Due to bandwidth limitation or insufficient device performance, data transmission may encounter delay or packet loss, which not only reduces the efficiency of key distribution, but also may cause security risks. More seriously, the congestion problem in quantum network directly affects the ability of key generation and distribution, thereby threatening the stability of the entire communication service.

[0004] Most of the existing QKD network routing selection is based on static network conditions, which is difficult to use in dynamic network environment, so as to effectively solve the congestion problem in QKD network. SUMMARY

[0005] The purpose of the present application is to provide a network congestion prediction method, device, electronic equipment and storage medium to predict the network congestion of quantum cryptography network and improve the performance and reliability of quantum cryptography network in view of the deficiencies in the prior art.

[0006] To achieve the above purpose, the technical solutions adopted by the embodiments of the present application are as follows:

[0007] In a first aspect, the embodiments of the present application provide a network congestion prediction method, which comprises:

[0008] Obtaining real-time network state data of multiple dimensions of quantum cryptography network, wherein the real-time network state data of multiple dimensions at least includes part or all of network delay, bandwidth utilization, packet loss rate, key pool remaining amount, key generation rate and key consumption rate;

[0009] Using a pre-trained hybrid prediction model of each dimension to predict the real-time network state data of each dimension, to obtain first network state data and second network state data of each dimension at the prediction time, wherein the first network state data and the second network state data are respectively the prediction results output by the first prediction model and the second prediction model in the hybrid prediction model;

[0010] According to the first network state data and the second network state data of each dimension at the prediction moment, target network state data of each dimension at the prediction moment is calculated;

[0011] According to the target network state data of multiple dimensions at the prediction moment, a congestion probability of the quantum cryptography network at the prediction moment is determined.

[0012] Optionally, the modeling step of the first prediction model in the hybrid prediction model of each dimension is as follows:

[0013] Obtain historical network state data of multiple historical moments of each dimension;

[0014] Differentially process the historical network state data of multiple historical moments of each dimension to obtain a stationary network state data sequence of each dimension, and determine a difference order of each dimension;

[0015] According to the stationary network state data sequence of each dimension, an autocorrelation function and a partial autocorrelation function of each dimension are calculated;

[0016] According to the autocorrelation function and the partial autocorrelation function of each dimension, an autoregressive order and a moving average order of each dimension are determined;

[0017] According to the difference order, the autoregressive order and the moving average order of each dimension, a first prediction model of each dimension is established.

[0018] Optionally, the training step of the second prediction model in the hybrid prediction model of each dimension is as follows:

[0019] Feature extraction is performed on the historical network state data of multiple historical moments of each dimension to obtain historical network state features of each dimension;

[0020] The historical network state features of each dimension are predicted by using the first prediction model of each dimension to determine first prediction data of a historical prediction moment;

[0021] According to the first prediction data and the historical network state data of the historical prediction moment, a first prediction data residual error is calculated;

[0022] The historical network state features of each dimension and the first prediction data residual error are predicted by using a preset long short-term memory network model to determine second prediction data of the historical prediction moment;

[0023] According to the second prediction data and the historical network state data of the historical prediction moment, a second prediction data residual error is calculated;

[0024] According to the second prediction data residual, the preset long short-term memory network model is optimized to obtain the second prediction model.

[0025] Optionally, the calculation of the autocorrelation function of each dimension according to the stationary network state data sequence of each dimension comprises:

[0026] According to the stationary network state data sequence of each dimension and a preset time lag parameter, the autocorrelation function of each dimension is calculated.

[0027] Optionally, the feature extraction of the historical network state data of a plurality of historical time points of each dimension comprises:

[0028] The periodic feature extraction of the historical network state data of a plurality of historical time points of each dimension is performed to determine the historical network periodic feature of each dimension.

[0029] The statistical feature extraction of the historical network state data of a plurality of historical time points of each dimension is performed to determine the historical network statistical feature of each dimension.

[0030] Optionally, the determination of the network congestion probability at the prediction time point according to the target network state data of the plurality of dimensions at the prediction time point comprises:

[0031] According to the target network state data of each dimension at the prediction time point and the mapping function of each dimension, the congestion influence factor of each dimension is calculated.

[0032] According to the congestion influence factors of the plurality of dimensions, the network congestion probability at the prediction time point is determined.

[0033] Optionally, the calculation of the congestion influence factor of each dimension according to the target network state data of each dimension at the prediction time point and the mapping function of each dimension comprises:

[0034] According to the predicted network delay at the prediction time point and the positive correlation mapping function of the network delay, the congestion influence factor of the network delay is calculated.

[0035] According to the predicted bandwidth usage at the prediction time point and the positive correlation mapping function of the bandwidth usage, the congestion influence factor of the bandwidth usage is calculated.

[0036] According to the predicted packet loss rate at the prediction time point and the positive correlation mapping function of the packet loss rate, the congestion influence factor of the packet loss rate is calculated.

[0037] a congestion influence factor of the key pool residual quantity is calculated according to a predicted key pool residual quantity at the predicted moment and a negative correlation mapping function of the key pool residual quantity;

[0038] a congestion influence factor of the key generation rate is calculated according to a predicted key generation rate at the predicted moment and a negative correlation mapping function of the key generation rate;

[0039] a congestion influence factor of the key consumption rate is calculated according to a predicted key consumption rate at the predicted moment and a positive correlation mapping function of the key consumption rate.

[0040] In a second aspect, the embodiments of the present application further provide a network congestion prediction device, the device comprising:

[0041] a data acquisition module configured to acquire real-time network state data of multiple dimensions of a quantum cryptography network, the real-time network state data of the multiple dimensions comprising at least part or all of network delay, bandwidth usage, packet loss rate, key pool residual quantity, key generation rate and key consumption rate;

[0042] a model prediction module configured to predict each dimension of real-time network state data by using a pre-trained hybrid prediction model of each dimension, to obtain first network state data and second network state data of each dimension at a predicted moment, the first network state data and the second network state data being prediction results output by a first prediction model and a second prediction model in the hybrid prediction model respectively;

[0043] a data calculation module configured to calculate target network state data of each dimension at the predicted moment according to the first network state data and the second network state data of each dimension at the predicted moment;

[0044] a probability calculation module configured to determine a congestion probability of the quantum cryptography network at the predicted moment according to the target network state data of the multiple dimensions at the predicted moment.

[0045] Optionally, a first prediction model modeling module is configured to acquire historical network state data of multiple historical moments of each dimension; perform differential processing on the historical network state data of multiple historical moments of each dimension to obtain a stationary network state data sequence of each dimension and determine a differential order of each dimension; calculate an autocorrelation function and a partial autocorrelation function of each dimension according to the stationary network state data sequence of each dimension; determine an autoregressive order and a moving average order of each dimension according to the autocorrelation function and the partial autocorrelation function of each dimension; and establish a first prediction model of each dimension according to the differential order, the autoregressive order and the moving average order of each dimension.

[0046] Optionally, the second prediction model training module is configured to perform feature extraction on the historical network state data of the multiple historical time points of each dimension to obtain historical network state features of each dimension; perform prediction on the historical network state features of each dimension by using the first prediction model of each dimension to determine first prediction data of a historical prediction time point; calculate first prediction data residuals according to the first prediction data and the historical network state data of the historical prediction time point; perform prediction on the historical network state features of each dimension and the first prediction data residuals by using a preset long short-term memory network model to determine second prediction data of the historical prediction time point; calculate second prediction data residuals according to the second prediction data and the historical network state data of the historical prediction time point; and optimize the preset long short-term memory network model according to the second prediction data residuals to obtain the second prediction model.

[0047] Optionally, the first prediction model modeling module is further configured to calculate an autocorrelation function of each dimension according to the stationary network state data sequence of each dimension and a preset time lag parameter.

[0048] Optionally, the second prediction model training module is further configured to perform periodic feature extraction on the historical network state data of the multiple historical time points of each dimension to determine historical network periodic features of each dimension; and perform statistical feature extraction on the historical network state data of the multiple historical time points of each dimension to determine historical network statistical features of each dimension.

[0049] Optionally, the probability calculation module is specifically configured to calculate congestion influence factors of each dimension according to the target network state data of each dimension at the prediction time point and the mapping function of each dimension; and determine a network congestion probability at the prediction time point according to the congestion influence factors of the multiple dimensions.

[0050] Optionally, the probability calculation module is further configured to calculate, according to a positive correlation mapping function of the predicted network delay at the prediction time and the network delay, a congestion influence factor of the network delay; calculate, according to a positive correlation mapping function of the predicted bandwidth usage at the prediction time and the bandwidth usage, a congestion influence factor of the bandwidth usage; calculate, according to a positive correlation mapping function of the predicted packet loss rate at the prediction time and the packet loss rate, a congestion influence factor of the packet loss rate; calculate, according to a negative correlation mapping function of the predicted key pool remaining amount at the prediction time and the key pool remaining amount, a congestion influence factor of the key pool remaining amount; calculate, according to a negative correlation mapping function of the predicted key generation rate at the prediction time and the key generation rate, a congestion influence factor of the key generation rate; and calculate, according to a positive correlation mapping function of the predicted key consumption rate at the prediction time and the key consumption rate, a congestion influence factor of the key consumption rate.

[0051] In a third aspect, an electronic device is provided, including a processor, a storage medium, and a bus. The storage medium stores program instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium through the bus. The processor executes the program instructions to perform the steps of the network congestion prediction method according to any one of the first aspect.

[0052] In a fourth aspect, a computer-readable storage medium is provided. The storage medium stores a computer program. When the computer program is run by a processor, the steps of the network congestion prediction method according to any one of the first aspect are performed.

[0053] The present application has the following beneficial effects:

[0054] The network congestion prediction method, device, electronic device, and storage medium provided by the present application can predict the real-time network state data of the quantum cryptography network in multiple dimensions based on a pre-trained hybrid prediction model, obtain first network state data and second network state data in multiple dimensions, and predict the linear and nonlinear changes of the real-time network state data through the first network state data and the second network state data, respectively. The target network state data calculated based on the first network state data and the second network state data can accurately reflect the network state changes of the quantum cryptography network in different dimensions, so that the congestion probability of the quantum cryptography network at the prediction time can be accurately predicted based on the target network state data in multiple dimensions, the prediction accuracy is ensured, the routing can be dynamically adjusted based on the congestion probability, the ability of the quantum cryptography network to cope with the load environment is improved, and the key transmission problem caused by congestion is reduced. In addition, by identifying and avoiding the congestion problem in advance, the delay and failure rate can be reduced, and the key distribution process can be smooth. BRIEF DESCRIPTION OF DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some of the embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0056] Figure 1 Flowchart of the network congestion prediction method provided by the embodiments of the present application Figure One

[0057] Figure 2 Flowchart of the network congestion prediction method provided by the embodiments of the present application Figure Two

[0058] Figure 3 Autocorrelation graph of the network delay dimension provided by the embodiments of the present application

[0059] Figure 4 Flowchart of the network congestion prediction method provided by the embodiments of the present application Figure Three

[0060] Figure 5 Schematic diagram of the hybrid prediction model provided by the embodiments of the present application

[0061] Figure 6 Prediction comparison graph of the network delay provided by the embodiments of the present application

[0062] Figure 7 Flowchart of the network congestion prediction method provided by the embodiments of the present application Figure Four

[0063] Figure 8 Prediction graph of the network congestion probability provided by the embodiments of the present application

[0064] Figure 9 Structural schematic diagram of the network congestion prediction device provided by the embodiments of the present application

[0065] Figure 10 Schematic diagram of the electronic device provided by the embodiments of the present application DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solutions and advantages of the embodiments of the present application more clear, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, not all the embodiments.

[0067] ​​​​Therefore, the following detailed description of embodiments of the application provided in the accompanying drawings is not intended to limit the scope of the application as claimed, but merely represents selected embodiments of the application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor are within the scope of protection of the present application.

[0068] In addition, the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a particular order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus including a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products, or apparatuses.

[0069] It should be noted that the features in the embodiments of the present application can be combined with each other without conflict.

[0070] A quantum cryptography network is a secure communication network using quantum cryptography technology, which is jointly constructed by a classical communication network and a quantum key distribution network. The classical communication network uses quantum keys to implement data encryption and decryption and encrypted data transmission. The quantum key distribution network is composed of QKD terminals and quantum links, and is used for key distribution.

[0071] The network nodes of the quantum cryptography network include user nodes (UN) and relay nodes (RN). User terminals are mounted on user nodes. Two communicating user nodes are connected through relay nodes. Relay nodes are used to extend the key transmission distance to perform key transmission between two user terminals at a long distance. The links formed between relay nodes are relay links (RL), which implement the communication function between two relay nodes. A user node is essentially a relay node, but it is called a user node because it needs to mount a user terminal.

[0072] The network nodes of the quantum cryptography network are composed of classical communication devices and QKD devices. The classical communication devices perform data encryption and decryption and encrypted data transmission through the classical communication network. The QKD devices are used for key distribution. Generally, the keys are distributed between relay nodes and between relay nodes and user nodes through QKD devices.

[0073] Figure 1Flowchart of network congestion prediction method provided for embodiments of the present application Figure One As shown in Figure 1 The method can include:

[0074] S101, obtain real-time network state data of multiple dimensions of the quantum cryptography network, the real-time network state data of multiple dimensions at least including part or all of network latency, bandwidth usage, packet loss rate, key pool remaining amount, key generation rate, and key consumption rate.

[0075] In the embodiment, the real-time network state data of multiple dimensions of the quantum cryptography network is real-time network state data of multiple dimensions of relay nodes of the quantum cryptography network, including real-time network state data of a classical communication network and real-time network state data of a quantum key distribution network.

[0076] Network latency (Network Latency) is the time delay required for data to be transmitted between relay links of relay nodes, i.e., the time delay for data transmission between relay nodes through a classical communication network in the quantum cryptography network; bandwidth usage is the bandwidth usage of the classical communication network for data transmission between relay nodes; and packet loss rate is the loss rate of data packets for data transmission between relay nodes through the classical communication network.

[0077] Key pool remaining amount (Key Remaining) is the number of quantum keys that have not been used or can be further distributed through the quantum key distribution network by the relay nodes; key generation rate (Key Generation Rate) is the rate of generating new keys between relay nodes through negotiation; and key consumption rate (Key Consumption Rate) is the rate of using or distributing keys by the relay nodes.

[0078] It should be noted that the real-time network state data used for network congestion prediction can include part or all of the above-mentioned multiple dimensions of data, and can also include data of other dimensions in addition to the above-mentioned multiple dimensions of data, and the embodiments are not limited in this regard.

[0079] S102, use a pre-trained hybrid prediction model of each dimension to predict real-time network state data of each dimension to obtain first network state data and second network state data of each dimension at a prediction time, the first network state data and the second network state data being prediction results output by a first prediction model and a second prediction model in the hybrid prediction model, respectively.

[0080] In the embodiment, a hybrid prediction model is trained in advance for the network status data of each dimension, and the hybrid prediction model comprises a first prediction model and a second prediction model.

[0081] The first prediction model is used to capture the linear change trend of the network status data of each dimension, and the second prediction model is used to process the nonlinear change in the network status data of each dimension. The real-time network status data of each dimension is input into the hybrid prediction model of each dimension. The linear change trend of the real-time network status data of each dimension is predicted by the first prediction model of each dimension to obtain the first network status data under the linear change of the real-time network status data. The nonlinear change in the real-time network status data of each dimension is predicted by the second prediction model of each dimension to obtain the second network status data under the nonlinear change of the real-time network status data.

[0082] S103. The target network status data of each dimension at the prediction time is calculated according to the first network status data and the second network status data of each dimension at the prediction time.

[0083] In the embodiment, the linear first network status data and the nonlinear second network status data predicted by the hybrid prediction model are combined to obtain the target network status data of each dimension at the prediction time.

[0084] S104. The congestion probability of the quantum cryptography network at the prediction time is determined according to the target network status data of multiple dimensions at the prediction time.

[0085] In the embodiment, the congestion probability of the quantum cryptography network at the prediction time is calculated according to the predicted network delay, the predicted bandwidth usage rate, the predicted packet loss rate, the predicted key pool remaining amount, the predicted key generation rate and / or the predicted key consumption rate at the prediction time.

[0086] Further, according to the congestion probability at the prediction time, it can be determined whether to select a new network route for the quantum cryptography network. If the congestion probability at the prediction time indicates that there is a high probability that the quantum cryptography network will be congested at the prediction time, a new network route can be selected for the quantum cryptography network in advance to avoid network congestion of the quantum cryptography network.

[0087] In a possible implementation manner, Figure 2 The flowchart of the network congestion prediction method provided by the embodiment of the application is shown in Figure Two As shown in Figure 2 The modeling steps of the first prediction model in the hybrid prediction model of each dimension are as follows:

[0088] S201. Obtain historical network state data of each dimension at multiple historical time points.

[0089] In this embodiment, the network state detection module in the quantum cryptography network collects historical network state data of each dimension at multiple historical time points related to the network node state, and the historical network state data carries timestamp information. The historical network state data of each dimension at multiple historical time points constitutes a univariate target sequence.

[0090] In some embodiments, after collecting the historical network state data of each dimension at multiple historical time points, it is necessary to clean up abnormal data and noise.

[0091] Specifically, the standard score (Z-Score) method is used to process the historical network state data at each historical time point to determine whether the historical network state data at each historical time point is an abnormal value. If it is an abnormal value, it is removed. For missing data, linear interpolation and polynomial interpolation can be used for supplementation.

[0092] For example, the formula of the standard score (Z-Score) method is as follows:

[0093]

[0094] where X is the historical network state data at each historical time point, μ X is the mean of the historical state data at multiple historical time points, and σ X is the standard deviation at multiple historical time points. When |Z| is greater than a preset value, for example, 3, it is determined that the corresponding historical network state data is abnormal data.

[0095] S202. Difference process the historical network state data of each dimension at multiple historical time points to obtain a stationary network state data sequence of each dimension, and determine the difference order of each dimension.

[0096] In this embodiment, an autoregressive moving average model (ARIMA) is used to construct a first prediction model.

[0097] The modeling process of the ARIMA model is a process of determining the difference order d, the autoregressive order p, and the moving average order q of the ARIMA model.

[0098] In the process of determining the difference order d, it is judged whether the historical network state data of multiple historical moments of each dimension is stationary. If the historical network state data of multiple historical moments of each dimension is stationary, the difference order of each dimension is determined to be 0. If the historical network state data of multiple historical moments of each dimension is not stationary, the historical network state data of multiple historical moments of each dimension is subjected to difference processing, and the difference order d is determined to change the historical network state data of multiple historical moments of each dimension from not stationary to stationary.

[0099] After each first-order difference processing, the difference data is subjected to a stationary detection. If it is stationary, the difference order is determined. If it is not stationary, the difference order is increased until the difference data is stationary, and the difference order is obtained.

[0100] For example, the first-order difference processing is to subtract the historical network state data of the first historical moment from the historical network state data of the second historical moment, as the first-order difference data of the first historical moment. The second-order difference processing is to subtract the difference between the historical network state data of the second historical moment and the historical network state data of the first historical moment from the difference between the historical network state data of the third historical moment and the historical network state data of the second historical moment, as the second-order difference data of the first historical moment.

[0101] S203, calculating the autocorrelation function and the partial autocorrelation function of each dimension according to the stationary network state data sequence of each dimension.

[0102] In this embodiment, the stationary network state sequence of each dimension includes the stationary network state data of multiple historical moments. The correlation between the stationary network state data of different historical moments in the stationary network state sequence of each dimension is analyzed to determine the autocorrelation function (ACF) and the partial autocorrelation function (PACF) of each dimension.

[0103] For example, for the first historical moment and the fourth historical moment, the autocorrelation function analyzes the correlation between the historical network state data of the first historical moment and the historical network state data of the fourth historical moment, which includes the influence of the historical network state data of the second historical moment and the historical network state data of the third historical moment. The partial autocorrelation function analyzes the correlation between the historical network state data of the first historical moment and the historical network state data of the fourth historical moment, which excludes the influence of the historical network state data of the second historical moment and the historical network state data of the third historical moment.

[0104] S204, determining the autoregressive order and the moving average order of each dimension according to the autocorrelation function and the partial autocorrelation function of each dimension.

[0105] In this embodiment, an autocorrelation plot is generated based on the autocorrelation function of each dimension, and the autoregression order p is determined from the autocorrelation plot. A partial autocorrelation plot is generated based on the partial autocorrelation function of each dimension, and the moving average order q is determined from the partial autocorrelation plot.

[0106] S205. Based on the difference order, autoregression order, and moving average order of each dimension, establish the first prediction model for each dimension.

[0107] In this embodiment, after determining the difference order d, autoregression order p, and moving average order q of the ARIMA model, the modeling of the ARIMA model is completed, and the first prediction model for each dimension is obtained.

[0108] In some embodiments, the ARIMA model is used to predict the network state data at historical prediction times. Based on the difference between the predicted data and the actual detected historical network state, the autoregressive order p and the moving average order q are modified to improve the accuracy of the first prediction model.

[0109] In one possible implementation, the process of calculating the autocorrelation function of each dimension based on the stationary network state data sequence of each dimension in step S203 may include:

[0110] Based on the stationary network state data sequence for each dimension and the preset time lag parameter, calculate the autocorrelation function for each dimension.

[0111] In this embodiment, the time lag parameter of the stationary network state data sequence is determined, and a correlation analysis is performed on the stationary network state data at the current moment and the stationary network state data at moments after the time lag parameter at the current moment based on the time lag parameter to determine the autocorrelation function of each dimension.

[0112] In some embodiments, if the historical network state data for each dimension changes periodically, and the preset time lag parameter is a fixed value, then the change period is determined as the preset time lag parameter. For example, when analyzing historical network state data for each day or each week, the preset time lag parameter is one day or one week.

[0113] If the historical network state data for each dimension is dynamically changing, and its preset time lag parameter is a dynamic value, the correlation between historical network state data can be analyzed from the autocorrelation graph of the autocorrelation function corresponding to different time lag parameters by continuously adjusting the time lag parameter, thereby determining the autoregression order p.

[0114] For example, the formula for calculating the autocorrelation function is shown below:

[0115]

[0116] Where τ is the time lag parameter and T is the total time length of multiple historical moments.

[0117] Example, Figure 3 The autocorrelation graph for the network latency dimension provided in the embodiments of this application is as follows: Figure 3 As shown, after plotting the autocorrelation function, the autocorrelation coefficient in the autocorrelation graph decays exponentially, and the autocorrelation function has a tail, indicating that the autoregressive order is 0. If the autocorrelation function is truncated, the autoregressive order is determined based on the truncation order.

[0118] In one possible implementation, Figure 4 A flowchart illustrating the network congestion prediction method provided in this application embodiment. Figure Three ,like Figure 4 As shown, the training steps for the second prediction model in the hybrid prediction model for each dimension are as follows:

[0119] S301. Extract features from the historical network state data at multiple historical moments for each dimension to obtain the historical network state features for each dimension.

[0120] In this embodiment, feature extraction is performed on the changing patterns and distribution of historical network state data at multiple historical moments for each dimension to obtain the historical network state features for each dimension.

[0121] In some embodiments, S301 above extracts features from historical network state data at multiple historical moments for each dimension to obtain historical network state features for each dimension, including:

[0122] Periodic features are extracted from historical network state data at multiple historical moments for each dimension to determine the periodic features of the historical network in each dimension; statistical features are extracted from historical network state data at multiple historical moments for each dimension to determine the statistical features of the historical network in each dimension.

[0123] In this embodiment, the network state data at multiple historical moments for each dimension are analyzed along the time dimension to determine the time information corresponding to the recurring states in the real-time network state data for each dimension, thereby identifying the historical network periodic characteristics for each dimension. Examples include daily traffic peaks and weekly usage habits.

[0124] In some embodiments, the following sine and cosine functions can be used to convert the time information of multiple historical moments t into periodic information, determine the period length T, analyze the historical network state data within the period length T, and determine the periodic characteristics of the historical network.

[0125]

[0126] For example, the cycle length can be days, weeks, months, etc.

[0127] Combining the historical network state data and the historical network periodicity features of multiple historical moments enables the model to capture time-related change trends in the quantum cryptography network and enhance the prediction capability of the periodic congestion risk.

[0128] The network state data of multiple historical moments for each dimension is counted to determine historical network statistical features, so as to evaluate the overall distribution of the network state data of multiple historical moments. The historical network statistical features can be one or more of the mean, variance, maximum value, and minimum value.

[0129] S302, the historical network state features of each dimension are predicted by using the first prediction model of each dimension to determine the first prediction data of the historical prediction moment.

[0130] In this embodiment, the historical network state features of each dimension are input into the first prediction model of each dimension, and the historical network state features of each dimension are linear trend modeled by the first prediction model of each dimension to output the first prediction data of the historical prediction moment.

[0131] In some embodiments, the historical network periodicity features and the historical network statistical features of each dimension are input into the first prediction model of each dimension, and the historical network periodicity features and the historical network statistical features of each dimension are linear trend modeled by the first prediction model of each dimension to output the first prediction data of the historical prediction moment.

[0132] S303, according to the first prediction data and the historical network state data of the historical prediction moment, the first prediction data residual is calculated.

[0133] In this embodiment, the historical network state data of the historical prediction moment is obtained, the difference between the historical network state data of the historical prediction moment and the first prediction data is calculated, and the first prediction data residual is extracted.

[0134] For example, the residual calculation formula is:

[0135]

[0136] Wherein, X(t) is the historical network state data of the historical prediction moment t, is the first prediction data of the historical prediction moment t. The first prediction data residual reflects the nonlinear component in the historical network state data of multiple historical moments.

[0137] In some embodiments, the historical network state feature corresponding to the historical network state data of the plurality of historical time points in different historical time periods is respectively input into the first prediction model, and first prediction data of a plurality of historical prediction time points is output. The first prediction data residual sequence is constructed according to the difference between the historical network state data of the plurality of historical prediction time points and the first prediction data of the corresponding historical prediction time point.

[0138] In S304, the preset long short-term memory network model is used to predict the historical network state feature of each dimension and the first prediction data residual, and second prediction data of the historical prediction time point is determined.

[0139] In this embodiment, the long short-term memory network model (Long Short-Term Memory, LSTM) is a kind of time recurrent neural network. The historical network state feature of each dimension and the first prediction data residual of the historical prediction time point are input into the LSTM model, the nonlinear time dependence relationship in the LSTM model is learned, the nonlinear trend of the network state data of each dimension is predicted, and the second prediction data of the historical prediction time point is output.

[0140] In S305, the second prediction data residual is calculated according to the second prediction data and the historical network state data of the historical prediction time point.

[0141] In this embodiment, the historical network state data of the historical prediction time point is obtained, and the difference between the historical network state data of the historical prediction time point and the second prediction data is calculated to determine the second prediction data residual.

[0142] In S306, the preset long short-term memory network model is optimized according to the second prediction data residual, and the second prediction model is obtained.

[0143] In this embodiment, the parameters of the LSTM model are optimized according to the second prediction data residual, and the LSTM model is iterated multiple times. After the number of iterations reaches a preset number or the second prediction data residual converges, the training of the LSTM model is completed, and the second prediction model is obtained.

[0144] In the example, Figure 5 The schematic diagram of the hybrid prediction model provided in the embodiment of the present application is shown in Figure 5 As shown in the figure, the univariate target sequence, i.e. the historical network state data of a plurality of historical time points, is preprocessed and input into the ARIMA model. The ARIMA model is modeled as a first prediction model. The first prediction model is used to predict the historical network periodicity feature and the historical network statistical feature, and output linear first network state data.

[0145] Based on the first network state data output by the ARIMA model, a residual sequence is constructed. The residual sequence and historical network state data are divided into datasets to determine the training set and test set. The parameters of the LSTM model are optimized to obtain the second prediction model. The second prediction model is used to predict the periodic characteristics and statistical characteristics of the historical network, and output nonlinear second network state data.

[0146] The prediction results of the first prediction model and the second prediction model, i.e., the first network state data and the second network state data, are combined to obtain the target network state data.

[0147] For example, the formula for calculating the target network state data can be expressed as:

[0148]

[0149] Furthermore, in addition to inputting real-time network state data into the hybrid prediction model to predict network state data for the next time period, the model parameters need to be updated periodically to ensure that the hybrid prediction model always reflects the latest network state.

[0150] Specifically, at preset time intervals, the latest real-time network state data is added to the training set, while the oldest historical network state data is removed, keeping the training set size constant. The model parameters are then retrained using the updated training set to ensure the model can adapt to new data distributions and trends. The trained model is then evaluated using a validation set to verify whether its predictive performance meets expectations, thus ensuring the effectiveness and accuracy of the model updates.

[0151] Example, Figure 6 A comparison chart of network latency predictions provided in the embodiments of this application, such as... Figure 6 As shown, the solid line represents the actual monitored network latency, and the dashed line represents the predicted network latency. It can be seen that the network latency predicted by the hybrid prediction model based on ARIMA and LSTM has high accuracy.

[0152] In one possible implementation, Figure 7 A flowchart illustrating the network congestion prediction method provided in this application embodiment. Figure Four ,like Figure 7 As shown, the process of determining the network congestion probability at the predicted time based on multiple dimensions of target network state data at the predicted time in S104 may include:

[0153] S401. Based on the target network state data for each dimension at the predicted time and the mapping function for each dimension, calculate the congestion impact factor for each dimension.

[0154] S402, determine the network congestion probability at the prediction time according to the congestion influence factors of multiple dimensions.

[0155] In this embodiment, in order to reflect the actual network state of the quantum cryptography network, it is necessary to map the network state data of each dimension to calculate the network congestion probability.

[0156] The network state data of each dimension has a corresponding mapping function f i (X(t)), for mapping the target network state data of each dimension to the congestion influence factor of the network congestion probability, according to the mapping function of each dimension, the target network state data of each dimension is calculated, the congestion influence factor of each dimension is determined, and the congestion influence factors of multiple dimensions are weighted and calculated to obtain the network congestion probability at the prediction time.

[0157] In some embodiments, the process of calculating the congestion influence factor of each dimension according to the target network state data of each dimension at the prediction time and the mapping function of each dimension in S401 can include:

[0158] According to the predicted network delay at the prediction time and the positive correlation mapping function of the network delay, the congestion influence factor of the network delay is calculated.

[0159] According to the predicted bandwidth usage at the prediction time and the positive correlation mapping function of the bandwidth usage, the congestion influence factor of the bandwidth usage is calculated.

[0160] According to the predicted packet loss rate at the prediction time and the positive correlation mapping function of the packet loss rate, the congestion influence factor of the packet loss rate is calculated.

[0161] According to the predicted key pool remaining amount at the prediction time and the negative correlation mapping function of the key pool remaining amount, the congestion influence factor of the key pool remaining amount is calculated.

[0162] According to the predicted key generation rate at the prediction time and the negative correlation mapping function of the key generation rate, the congestion influence factor of the key generation rate is calculated.

[0163] According to the predicted key consumption rate at the prediction time and the positive correlation mapping function of the key consumption rate, the congestion influence factor of the key consumption rate is calculated.

[0164] In this embodiment, the greater the network delay L(t), the greater the congestion probability, i.e. the mapping function of the network delay on the network congestion probability is a positive correlation mapping function. The positive correlation mapping function of the network delay L(t) can be set as the following linear relationship:

[0165]

[0166] The higher the bandwidth usage rate B(t) is, the greater the congestion probability is, that is, the mapping function of the influence of the bandwidth usage rate on the network congestion probability is a positive correlation mapping function. The positive correlation mapping function of the bandwidth usage rate B(t) can be set as the following linear relationship:

[0167]

[0168] The higher the packet loss rate P(t) is, the greater the congestion probability is, that is, the mapping function of the influence of the packet loss rate on the network congestion probability is a positive correlation mapping function. The positive correlation mapping function of the packet loss rate P(t) can be set as the following linear relationship:

[0169]

[0170] The smaller the remaining amount of the key pool K r (t) is, the greater the congestion probability is, that is, the mapping function of the influence of the remaining amount of the key pool on the network congestion probability is a negative correlation mapping function. The negative correlation mapping function of the remaining amount of the key pool K r (t) can be set as the following linear relationship:

[0171]

[0172] The slower the key generation rate K g (t) is, the greater the congestion probability is, that is, the mapping function of the influence of the key generation rate on the network congestion probability is a negative correlation mapping function. The negative correlation mapping function of the key generation rate K g (t) can be set as the following linear relationship:

[0173]

[0174] The faster the key consumption rate K c (t) is, the greater the congestion probability is, that is, the mapping function of the influence of the key consumption rate on the network congestion probability is a positive correlation mapping function. The positive correlation mapping function of the key consumption rate Kc(t) can be set as the following linear relationship:

[0175]

[0176] The final calculation formula of the congestion probability can be represented as:

[0177]

[0178] wherein, wi is the congestion weight of each dimension, the total value of all the congestion weights is 1, f i (X(t)) is the mapping function of each dimension, and N is the number of dimensions, which is 6 in this embodiment.

[0179] An example of the prediction diagram of the network congestion probability provided by the embodiment of the present application is shown in FIG. 1. Figure 8 An example of the prediction diagram of the network congestion probability provided by the embodiment of the present application is shown in FIG. 1.Figure 8 As shown, the network congestion prediction method provided by the embodiment can be used to predict the network congestion probability of the quantum cryptography network at each moment.

[0180] The network congestion prediction method provided by the above embodiment can be used to predict the real-time network state data of the quantum cryptography network in multiple dimensions based on the pre-trained hybrid prediction model, to obtain first network state data and second network state data in multiple dimensions, and to predict the linear change and nonlinear change of the real-time network state data by the first network state data and the second network state data. The target network state data calculated based on the first network state data and the second network state data can accurately reflect the network state change of the quantum cryptography network in different dimensions, so that the congestion probability of the quantum cryptography network at the prediction moment can be accurately predicted based on the target network state data in multiple dimensions, the prediction accuracy is ensured, the routing can be dynamically adjusted based on the congestion probability, the ability of the quantum cryptography network to cope with the load environment is improved, and the key transmission problem caused by congestion is reduced. In addition, by identifying and avoiding the congestion problem in advance, the delay and failure rate can be reduced, and the key distribution process can be ensured to be smooth.

[0181] Based on the above method embodiment, the embodiment of the present application further provides a network congestion prediction device. Figure 9 The structure diagram of the network congestion prediction device provided by the embodiment of the present application is shown in Figure 9 As shown, the device can include:

[0182] The data acquisition module 501 is configured to acquire real-time network state data of the quantum cryptography network in multiple dimensions, and the real-time network state data in multiple dimensions at least includes some or all of network delay, bandwidth usage, packet loss rate, key pool remaining amount, key generation rate, and key consumption rate.

[0183] The model prediction module 502 is configured to predict the real-time network state data in each dimension by using a pre-trained hybrid prediction model in each dimension, to obtain first network state data and second network state data in each dimension at the prediction moment, and the first network state data and the second network state data are prediction results output by a first prediction model and a second prediction model in the hybrid prediction model.

[0184] The data calculation module 503 is configured to calculate target network state data in each dimension at the prediction moment based on the first network state data and the second network state data in each dimension at the prediction moment.

[0185] The probability calculation module 504 is configured to determine the congestion probability of the quantum cryptography network at the prediction moment based on the target network state data in multiple dimensions at the prediction moment.

[0186] Optionally, the first prediction model modeling module is configured to: obtain historical network state data of a plurality of historical time points of each dimension; perform difference processing on the historical network state data of the plurality of historical time points of each dimension to obtain a stationary network state data sequence of each dimension, and determine a difference order of each dimension; calculate an autocorrelation function and a partial autocorrelation function of each dimension according to the stationary network state data sequence of each dimension; determine an autoregressive order and a moving average order of each dimension according to the autocorrelation function and the partial autocorrelation function of each dimension; and establish the first prediction model of each dimension according to the difference order, the autoregressive order and the moving average order of each dimension.

[0187] Optionally, the second prediction model training module is configured to: perform feature extraction on the historical network state data of the plurality of historical time points of each dimension to obtain historical network state features of each dimension; perform prediction on the historical network state features of each dimension by using the first prediction model of each dimension to determine first prediction data of a historical prediction time point; calculate a first prediction data residual according to the first prediction data and the historical network state data of the historical prediction time point; perform prediction on the historical network state features of each dimension and the first prediction data residual by using a preset long short-term memory network model to determine second prediction data of the historical prediction time point; calculate a second prediction data residual according to the second prediction data and the historical network state data of the historical prediction time point; and optimize the preset long short-term memory network model according to the second prediction data residual to obtain the second prediction model.

[0188] Optionally, the first prediction model modeling module is further configured to calculate the autocorrelation function of each dimension according to the stationary network state data sequence of each dimension and a preset time lag parameter.

[0189] Optionally, the second prediction model training module is further configured to: perform periodic feature extraction on the historical network state data of the plurality of historical time points of each dimension to determine historical network periodic features of each dimension; and perform statistical feature extraction on the historical network state data of the plurality of historical time points of each dimension to determine historical network statistical features of each dimension.

[0190] Optionally, the probability calculation module 504 is specifically configured to: calculate a congestion influence factor of each dimension according to the target network state data of each dimension at the prediction time point and the mapping function of each dimension; and determine a network congestion probability at the prediction time point according to the congestion influence factors of the plurality of dimensions.

[0191] Optionally, the probability calculation module 504 is further configured to calculate the congestion impact factor of network latency based on the predicted network latency and the positive correlation mapping function of network latency at the prediction time; calculate the congestion impact factor of bandwidth utilization based on the predicted bandwidth utilization and the positive correlation mapping function of bandwidth utilization at the prediction time; calculate the congestion impact factor of packet loss rate based on the predicted packet loss rate and the positive correlation mapping function of packet loss rate at the prediction time; calculate the congestion impact factor of key pool remaining amount based on the negative correlation mapping function of key pool remaining amount at the prediction time; calculate the congestion impact factor of key generation rate based on the negative correlation mapping function of key generation rate and key generation rate at the prediction time; and calculate the congestion impact factor of key consumption rate based on the positive correlation mapping function of key consumption rate and key consumption rate at the prediction time.

[0192] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

[0193] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more microprocessors, or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).

[0194] Figure 10 A schematic diagram of the electronic device provided in the embodiments of this application, such as... Figure 10 As shown, the electronic device 600 may include a processor 601, a storage medium 602, and a bus. The storage medium 602 stores program instructions executable by the processor 601. When the electronic device 600 is running, the processor 601 communicates with the storage medium 602 via the bus, and the processor 601 executes the program instructions to perform the above-described method embodiment. The specific implementation and technical effects are similar and will not be described in detail here.

[0195] Optionally, this application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, is used to perform the above-described method embodiments.

[0196] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented by other manners. For example, the apparatus embodiments described above are only illustrative, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some interfaces, apparatuses or units, and can be electrical, mechanical or other forms.

[0197] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place or distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0198] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of hardware plus software functional unit.

[0199] The integrated unit realized in the form of software functional unit can be stored in a computer readable storage medium. The software functional unit stored in a storage medium includes a plurality of instructions for making a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (English: processor) execute part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes: U disk, mobile hard disk, read-only memory (English: Read-Only Memory, abbreviated as: ROM), random access memory (English: Random Access Memory, abbreviated as: RAM), magnetic disk or optical disk and various program code storage media.

[0200] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A network congestion prediction method, characterized in that, The method includes: To obtain real-time network status data of a quantum cryptography network from multiple dimensions, wherein the real-time network status data from multiple dimensions includes at least some or all of the following: network latency, bandwidth utilization, packet loss rate, remaining key pool amount, key generation rate, and key consumption rate. The real-time network state data of each dimension is predicted by a pre-trained hybrid prediction model for each dimension, and the first network state data and the second network state data of each dimension at the prediction time are obtained. The first network state data and the second network state data are the prediction results output by the first prediction model and the second prediction model in the hybrid prediction model, respectively. Based on the first network state data and the second network state data of each dimension at the prediction time, calculate the target network state data of each dimension at the prediction time. Based on the target network state data of the multiple dimensions at the predicted time, determine the congestion probability of the quantum cryptography network at the predicted time; The training steps for the second prediction model in the hybrid prediction model for each dimension are as follows: Obtain historical network state data at multiple historical moments for each dimension; Feature extraction is performed on the historical network state data at multiple historical moments for each dimension to obtain the historical network state features for each dimension; The first prediction model for each dimension is used to predict the historical network state features of each dimension, and the first prediction data for the historical prediction time is determined. Calculate the residual of the first prediction data based on the first prediction data and the historical network state data at the historical prediction time; A preset long short-term memory network model is used to predict the historical network state features of each dimension and the residual of the first prediction data to determine the second prediction data at the historical prediction time. Calculate the residual of the second prediction data based on the second prediction data and the historical network state data at the historical prediction time; The preset long short-term memory network model is optimized based on the residual of the second prediction data to obtain the second prediction model.

2. The method as described in claim 1, characterized in that, The modeling steps for the first prediction model in the hybrid prediction model for each dimension are as follows: Obtain historical network state data at multiple historical moments for each dimension; Differential processing is performed on the historical network state data at multiple historical moments for each dimension to obtain a stationary network state data sequence for each dimension, and the difference order for each dimension is determined. Calculate the autocorrelation function and partial autocorrelation function for each dimension based on the stationary network state data sequence for each dimension; Based on the autocorrelation function and partial autocorrelation function of each dimension, determine the autoregression order and moving average order of each dimension; Based on the difference order, autoregression order, and moving average order of each dimension, a first prediction model is established for each dimension.

3. The method as described in claim 2, characterized in that, The step of calculating the autocorrelation function for each dimension based on the stationary network state data sequence for each dimension includes: The autocorrelation function of each dimension is calculated based on the stationary network state data sequence of each dimension and the preset time lag parameter.

4. The method as described in claim 1, characterized in that, The step of extracting features from historical network state data at multiple historical moments for each dimension to obtain historical network state features for each dimension includes: Periodic features are extracted from the historical network state data at multiple historical moments for each dimension to determine the historical network periodic features for each dimension. Statistical features are extracted from the historical network state data at multiple historical moments for each dimension to determine the historical network statistical features for each dimension.

5. The method as described in claim 1, characterized in that, Determining the network congestion probability at the predicted time based on the target network state data of the multiple dimensions at the predicted time includes: Based on the target network state data for each dimension at the predicted time and the mapping function for each dimension, calculate the congestion impact factor for each dimension; Based on the multiple congestion impact factors, the network congestion probability at the predicted time is determined.

6. The method as described in claim 5, characterized in that, The step of calculating the congestion impact factor for each dimension based on the target network state data for each dimension at the predicted time and the mapping function for each dimension includes: Calculate the congestion impact factor of the network latency based on the predicted network latency at the predicted time and the positive correlation mapping function of the network latency; Based on the predicted bandwidth utilization at the predicted time and the positive correlation mapping function of the bandwidth utilization, calculate the congestion impact factor of the bandwidth utilization; Calculate the congestion impact factor of the packet loss rate based on the predicted packet loss rate at the predicted time and the positive correlation mapping function of the packet loss rate; Calculate the congestion impact factor of the remaining key pool based on the predicted key pool remaining amount at the predicted time and the negative correlation mapping function of the remaining key pool amount; Calculate the congestion impact factor of the key generation rate based on the predicted key generation rate at the predicted time and the negative correlation mapping function of the key generation rate; The congestion impact factor of the key consumption rate is calculated based on the predicted key consumption rate at the predicted time and the positive correlation mapping function of the key consumption rate.

7. A network congestion prediction device, characterized in that, The device includes: The data acquisition module is used to acquire real-time network status data of the quantum cryptography network in multiple dimensions. The real-time network status data in multiple dimensions includes at least some or all of the following: network latency, bandwidth utilization, packet loss rate, key pool remaining amount, key generation rate, and key consumption rate. The model prediction module is used to predict the real-time network state data of each dimension using a pre-trained hybrid prediction model for each dimension, and to obtain the first network state data and the second network state data of each dimension at the prediction time. The first network state data and the second network state data are the prediction results output by the first prediction model and the second prediction model in the hybrid prediction model, respectively. The data calculation module is used to calculate the target network state data for each dimension at the prediction time based on the first network state data and the second network state data for each dimension at the prediction time. The probability calculation module is used to determine the congestion probability of the quantum cryptography network at the prediction time based on the target network state data of the multiple dimensions at the prediction time. The device further includes: The second prediction model training module is used to acquire historical network state data at multiple historical moments for each dimension; extract features from the historical network state data at multiple historical moments for each dimension to obtain historical network state features for each dimension; predict the historical network state features for each dimension using the first prediction model for each dimension to determine the first prediction data for the historical prediction moment; calculate the first prediction data residual based on the first prediction data and the historical network state data for the historical prediction moment; predict the historical network state features for each dimension and the first prediction data residual using a preset long short-term memory network model to determine the second prediction data for the historical prediction moment; calculate the second prediction data residual based on the second prediction data and the historical network state data for the historical prediction moment; and optimize the preset long short-term memory network model based on the second prediction data residual to obtain the second prediction model.

8. An electronic device, characterized in that, include: The device includes a processor, a storage medium, and a bus, wherein the storage medium stores program instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the program instructions to perform the steps of the network congestion prediction method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, performs the steps of the network congestion prediction method as described in any one of claims 1 to 6.

Images