A network traffic grey prediction method based on an adaptive feedback adjustment mechanism
By optimizing the GM(1,1) grey model through an adaptive feedback adjustment mechanism, the problems of poor accuracy and efficiency in network traffic prediction are solved, and high-precision and high-efficiency traffic prediction results are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING UNIV OF INFORMATION SCI & TECH
- Filing Date
- 2022-09-29
- Publication Date
- 2026-06-05
AI Technical Summary
Existing network traffic prediction methods struggle to achieve a good balance between accuracy and efficiency, and grey prediction algorithms are highly complex.
An adaptive feedback adjustment mechanism is adopted. By constructing a feedback correction function and window sliding prediction, the feedback factor is adjusted to optimize the GM(1,1) grey model, thereby achieving high-precision prediction of network traffic.
It improves traffic prediction accuracy, reduces prediction error, and enhances prediction efficiency. The average prediction time is approximately 0.65 seconds, which is 6.49% higher than the GM(1,1) model and 1.55% higher than the ewboGM(1,1) model.
Smart Images
Figure CN115567405B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of information engineering, and specifically relates to a gray prediction method for network traffic based on an adaptive feedback adjustment mechanism. Background Technology
[0002] With the rapid development of computer network technology, the number of network nodes and network applications has exploded, and the scale and complexity of network traffic have become increasingly large and complex. This has prompted the evolution of network communication architecture from the traditional "fixed & planned" network to the "sensing & control" intelligent network. In the "sensing & control" intelligent network communication architecture, key nodes will possess computing and storage capabilities. Due to computing capabilities, large-volume data is processed into smaller-volume data, and data content is compressed. Due to storage capabilities, the bursty characteristics of network traffic are weakened or enhanced, and traffic characteristics change accordingly. These changes will affect the distribution patterns and statistical characteristics of traffic data during the convergence of heterogeneous links in intelligent networks, causing changes in the related network traffic characteristic descriptions and affecting the accuracy of intelligent network traffic prediction. Currently, network traffic prediction methods are mainly divided into linear prediction and nonlinear prediction.
[0003] (1) Linear prediction
[0004] Linear prediction methods mainly include ARMA, ARIMA, and FARIMA. Duan Huaqiong et al. from Sichuan University proposed a combined prediction method based on the ARMA model, which realizes the combined prediction of network data using multiple linear models at different scales. Simulation results show that the mean prediction error of this method is within 10%. -3 The above linear prediction methods have high prediction accuracy. Tian Zhongda et al. from Shenyang University of Technology proposed a network traffic prediction model that compensates the autoregressive integral moving average (ARIMA) model for Gaussian process regression. Simulation results show that this method has higher prediction accuracy and smaller prediction error. Sun Qiang et al. from Beijing Jiaotong University proposed a novel railway data network traffic prediction method based on the FARIMA model. Simulation results show that this method is more accurate than the traditional ARMA model-based prediction method. While the above linear prediction methods have good prediction effects for stationary sequences, they are difficult to capture the implicit information in non-stationary sequences.
[0005] (2) Nonlinear prediction
[0006] Nonlinear prediction methods mainly include prediction methods related to artificial intelligence, such as machine learning and deep learning. Yingqi Li from Nanjing University of Posts and Telecommunications proposed a smoothed-assisted support vector machine (SSVM) model for predicting non-stationary video traffic. Experimental results show that the proposed SSVM model has significant advantages in prediction accuracy and statistical comparison. Hongsuk Yi from the Korea Institute of Scientific and Technical Information proposed a deep learning model for traffic prediction based on hyperparameter search. Experimental results show that this model has excellent traffic prediction performance. Although these methods have high prediction accuracy, the complexity of mathematical modeling and reliance on large sample data make the overhead in the traffic prediction process considerable.
[0007] While the aforementioned methods can effectively predict network traffic, they all fail to achieve a good balance between accuracy and efficiency. To address this issue, some domestic scholars have proposed methods for network traffic prediction based on grey prediction theory. To improve the prediction accuracy of grey models, some researchers have combined grey prediction models with Markov processes, calculating the one-step transition probability of current traffic to obtain the frequency of traffic increases, decreases, and stabilization at the next moment, thereby improving the accuracy of traffic prediction. However, the grey traffic prediction algorithm of this method has high complexity. Summary of the Invention
[0008] The purpose of this invention is to address the current limitations of traffic prediction methods in achieving a good balance between accuracy and efficiency. It aims to achieve high-precision prediction of network traffic while maintaining high prediction efficiency.
[0009] To achieve the above objectives, the present invention provides the following technical solution: a network traffic grey prediction method based on an adaptive feedback adjustment mechanism, which is based on a pre-constructed network topology and a continuous traffic sequence of the target network topology to be predicted. Perform the following steps S1 to S8 to obtain the target continuous flow sequence to be predicted. Predicted flow sequences;
[0010] Step S1: Process the flow sequence Normalization is performed to obtain a normalized flow sequence. ;
[0011] Step S2: Construct a model based on the correction factor Feedback correction function ;
[0012] Step S3: Initialize the feedback correction function Feedback factor ,set up The initial value is ;
[0013] Step S4: Based on the normalized flow sequence Feedback correction function and feedback correction function Feedback factor For normalized flow sequences A feedback correction function is applied to each element in the process. Obtain the preprocessed flow sequence ;
[0014] Step S5: Use a sliding window prediction method to... Implement GM(1,1) grey prediction and retain the windowed prediction values at each time step in sequence;
[0015] Step S6: Based on the window sliding prediction value at each time step S5, assign the window sliding prediction value at each time step to its corresponding value in... The flow rate values at the next time step are subtracted and compared to obtain the error of each flow rate sequence. Then, the error is predicted based on a preset threshold value. Three types of relationships between predicted and actual values are obtained: the difference between the predicted and actual values is approximately equal within a preset range, the predicted value is greater than the actual value, and the predicted value is less than the actual value.
[0016] Step S7: When the difference between the predicted value and the true value is approximately equal to a preset range, maintain the feedback factor for the window sliding prediction at the next time step. The size remains unchanged, that is Then return to step S4;
[0017] When the predicted value is greater than the actual value, based on the preset feedback factor Corrected gradient Update the correction factor size for the next time-step window sliding prediction to... Then return to step S4;
[0018] When the predicted value is less than the actual value, based on the preset feedback factor Corrected gradient Update the correction factor size for the next time-step window sliding prediction to... Then return to step S4;
[0019] Step S8: Inverse normalize the series of flow prediction results obtained by traversing the flow sequence through the sliding window to obtain the predicted flow sequence.
[0020] Furthermore, the aforementioned step S1 specifically involves: processing the flow sequence. Normalization is performed to obtain the normalized flow sequence. , .
[0021] Furthermore, in step S2 above, the correction factor is constructed according to the following formula. Feedback correction function :
[0022]
[0023] in, It is a feedback factor.
[0024] Furthermore, step S4 described above specifically involves generating a preprocessed flow sequence according to the following formula. :
[0025] .
[0026] Furthermore, the aforementioned step S5 includes the following sub-steps:
[0027] S5.1, For the flow sequence Initialize and obtain the initialized traffic sequence. ;
[0028] S5.2, The preset sliding prediction window size is m ,Will consecutive adjacent m Each flow sample is used as the initial sequence for each prediction. : ,in ;
[0029] S5.3 Calculation A single accumulation generates a sequence :
[0030] ;
[0031] in,
[0032] ;
[0033] S5.4 Calculate according to the following formula The nearest neighbor mean generation sequence :
[0034] ,
[0035] in,
[0036]
[0037] S5.5, Construct according to the following formula and The grey differential equation:
[0038]
[0039] in, The gray development coefficient, This is the gray action quantity;
[0040] S5.6 Construct the parameter vector of the grey differential equation from step S5.5. And solve it using the least squares method according to the following formula: ,
[0041] in,
[0042] ;
[0043] S5.7, obtained from the solution in step S5.6 , Calculate the time response function of the grey differential equation using the following formula. :
[0044]
[0045] S5.8, obtained from the solution in step S5.7 The prediction results are calculated using the following formula. :
[0046] .
[0047] Furthermore, the specific relationships between the three predicted values and the true values obtained in step S6 above are as follows:
[0048] S6.1 Judgment If the condition is met, then the difference between the predicted value and the actual value is approximately equal to the preset range; otherwise, proceed to step S6.2.
[0049] S6.2, Judgment If the condition is true, then the predicted value is greater than the actual value; otherwise, the predicted value is less than the actual value.
[0050] Furthermore, the aforementioned network topology includes at least one traffic sending node and one traffic receiving node, with a P2P communication connection between the traffic sending node and the traffic receiving node.
[0051] Furthermore, the aforementioned gray prediction method for network traffic based on an adaptive feedback adjustment mechanism uses Wireshark packet capture software to continuously capture the traffic to be predicted in the network topology.
[0052] The network traffic grey prediction method based on an adaptive feedback adjustment mechanism of the present invention, compared with the prior art, has the following technical advantages: improved traffic prediction accuracy. Compared to the GM(1,1) model prediction It improved by 6.49%, compared to the prediction of the ewboGM(1,1) model. The accuracy was improved by 1.55%. Under an Intel(R) Core(TM) i7-9700 CPU@3GHz processor, 32GB of memory, and a 64-bit Win10 operating system, the average prediction time of the algorithm was approximately 0.65 seconds. These results demonstrate that the proposed network traffic grey prediction method based on an adaptive feedback adjustment mechanism has advantages in both prediction accuracy and efficiency. Attached Figure Description
[0053] Figure 1 This is an algorithm flowchart of the method of the present invention.
[0054] Figure 2 This is the GM(1,1) grey prediction flowchart.
[0055] Figure 3 This is a diagram illustrating the principle of window sliding one step prediction.
[0056] Figure 4 This is a schematic diagram of the adaptive feedback adjustment mechanism.
[0057] Figure 5 This is a visualization of network traffic simulation results using PyViz based on NS3.
[0058] Figure 6 This is a graph of self-similar network traffic data obtained by the Wireshark packet capture software.
[0059] Figure 7 This is a comparison chart of the predicted and actual network traffic values using the method of this invention.
[0060] Figure 8 In the process of network traffic prediction, feedback factors The regulatory process.
[0061] Figure 9 There are three initial types. A comparison chart of the normalized absolute error of traffic prediction.
[0062] Figure 10 This is a comparison chart of the predicted and actual network traffic values based on the traditional GM(1,1) model.
[0063] Figure 11 This is a comparison chart of the predicted and actual network traffic values based on the ewboGM(1,1) model.
[0064] Figure 12 This is a comparison chart of the normalized absolute errors of the traffic flow predictions and actual values from the three models. Detailed Implementation
[0065] To better understand the technical content of the present invention, specific embodiments are described below in conjunction with the accompanying drawings.
[0066] In this invention, various aspects of the invention are described with reference to the accompanying drawings, in which numerous illustrative embodiments are shown. Embodiments of the invention are not limited to those depicted in the drawings. It should be understood that the invention is implemented through any of the various concepts and embodiments described above, as well as the concepts and embodiments described in detail below, because the concepts and embodiments disclosed herein are not limited to any particular implementation. Furthermore, some aspects of the invention disclosed may be used alone or in any suitable combination with other aspects of the invention disclosed.
[0067] Figure 1 The algorithm flowchart of the method of the present invention specifically includes the following steps:
[0068] S1, for the flow sequence Normalization is performed to prevent insufficient prediction accuracy due to excessive absolute error, resulting in the normalized flow sequence. Among them, .
[0069] S2, Constructing based on correction factors Feedback correction function For most systems, the data sequence is non-stationary, especially in intelligent networks where the burstiness of network traffic is more pronounced, and system shocks and disturbances are more significant, greatly reducing the accuracy of traffic prediction. To reduce system shocks and disturbances and improve traffic prediction accuracy, a feedback correction function is designed. , It consists of an exponentially weakened buffer operator and a feedback factor.
[0070] 1) The weakening buffer operator has the following theorem:
[0071] Theorem 1: Let It is a non-negative system behavior data sequence, and ,make
[0072]
[0073] Then when When the sequence is a monotonically increasing sequence, a monotonically decreasing sequence, or an oscillating sequence, All of these are weakened buffer operators. This is the exponential weakening buffer operator;
[0074] 2) Based on Theorem 1, construct a feedback correction function based on the feedback factor. ,have
[0075]
[0076] in, This is the feedback factor, whose function is to change by adjusting its own size. The intensity of its effect.
[0077] 3) Based on 2), prove It still belongs to the exponential weakened buffer operator. Since the flow sequence fluctuates with time, it belongs to the oscillating sequence. Based on Theorem 1, the relevant proof is as follows:
[0078] Proof 1: When When it is an oscillating sequence, let
[0079]
[0080] because
[0081] ,
[0082] so .
[0083] because
[0084]
[0085] so .
[0086] Therefore When it is an oscillating sequence, To weaken the buffer operator.
[0087] 4) Discussion and The relationship. Proof 1 shows that introducing back, It remains a weakened buffer operator, therefore its buffering properties are unchanged. According to proof 1, we have...
[0088]
[0089] at this time Satisfying a negative exponential function, when When it increases, Increase, and thus make Decrease; similarly, The reduction will make Increase.
[0090]
[0091] at this time Satisfies an exponential function, when When it increases, Similarly, it increases, thus making Decrease; similarly, The reduction will make Increase. Ultimately, we can conclude that when When it is an oscillating sequence, along with The value decreases as it increases, and increases as it decreases. This conclusion can be used as a basis for flow prediction. The basis for the adjustment.
[0092] S3, Initialize feedback correction function Feedback factor ,set up The initial value is ,make ;
[0093] S4. Based on the normalized flow sequence Feedback correction function and feedback correction function Feedback factor For normalized flow sequences A feedback correction function is applied to each element in the process. The preprocessed flow sequence is generated according to the following formula. :
[0094] .
[0095] S5. Employ a one-step sliding window prediction method for... The GM(1,1) grey prediction method is implemented, and its grey prediction process is as follows: Figure 2 As shown, the windowed prediction values at each time step are retained sequentially; to ensure the accuracy of the flow prediction, the prediction window size is set to 5, i.e. Five consecutive adjacent flow samples are used as the initial sequence for each prediction. :
[0096] .
[0097] Its window sliding one-step prediction principle is as follows: Figure 3 As shown, the prediction window slides continuously to the right to ensure continuous updates of traffic data, thereby achieving sliding prediction of the entire traffic sequence.
[0098] calculate A single accumulation generates a sequence ,have
[0099]
[0100] in,
[0101]
[0102] calculate The nearest neighbor mean generation sequence ,have
[0103]
[0104] in,
[0105]
[0106] Build and The grey differential equation has
[0107]
[0108] in, The gray development coefficient, This represents the gray action quantity.
[0109] Constructing the parameter vector of the grey differential equation And solve it using the least squares method, we have
[0110]
[0111] in,
[0112]
[0113] The solution obtained and Calculate the time response function of the grey differential equation. ,have
[0114]
[0115] The solution obtained Calculate the prediction results ,have
[0116]
[0117] S6. Based on the window sliding prediction value at each moment in step S5, classify the possible situations of traffic prediction, specifically: classify the window sliding prediction value at each moment with its corresponding... The flow rate value at the next time step is subtracted and compared to obtain the flow rate sequence error. Then, a preset flow rate prediction threshold error value is used. Three types of relationships between predicted and actual values are obtained: the difference between the predicted and actual values is approximately equal within a preset range, the predicted value is greater than the actual value, and the predicted value is less than the actual value.
[0118] set up Always The predicted flow rate at time is , The actual value of the flow at any given time is ,
[0119] S6.1 Judgment Whether it is true or not, if so, then it is considered true. ,show time Its effect intensity is moderate, therefore Unchanged, in order to maintain time The strength of the effect; determine if the predicted value is approximately equal to the true value; otherwise, proceed to step S6.2;
[0120] S6.2, Judgment Whether it is true or not, if it is true, it means that... time The effect intensity is too weak, therefore it needs to be reduced. To increase thereby enhancing time The strength of the effect determines whether the predicted value is greater than the true value; otherwise, the predicted value is less than the true value. ,show time The effect intensity is too high, therefore it needs to be increased. To reduce thus weakening time The intensity of its effect.
[0121] S7. Based on the relationship between the three predicted values and the actual values obtained in step S6, formulate the adjustment criteria for the adaptive feedback adjustment mechanism. The feedback adjustment principle is as follows: Figure 4 As shown.
[0122] When the predicted value is approximately equal to the actual value, maintain the feedback factor for the window sliding prediction at the next time step. The size remains unchanged, that is Then return to step S4;
[0123] When the predicted value is greater than the actual value, based on the preset feedback factor Corrected gradient Update the correction factor size for the next time-step window sliding prediction to... ,in Then return to step S4;
[0124] When the predicted value is less than the actual value, based on the preset feedback factor Corrected gradient Update the correction factor size for the next time-step window sliding prediction to... ,in Then return to step S4;
[0125] S8. The series of traffic prediction results obtained by sliding the window through the traffic sequence are denormalized to obtain the predicted traffic sequence.
[0126] This invention models and simulates network topology in multi-link aggregation scenarios using NS3. The simulated network traffic visualization results obtained through PyViz are as follows: Figure 5 As shown in the figure, the constructed network topology has 9 nodes, where nodes D1 to D8 represent traffic sending nodes and node R represents traffic receiving nodes. The nodes communicate via P2P links. Traffic was captured using Wireshark for 500 consecutive time points, resulting in the simulated traffic sequence shown below. Figure 6 As shown. From Figure 6 It can be seen that the network traffic obtained from the simulation is highly bursty, thus placing high demands on the prediction accuracy of the traffic prediction method.
[0127] Set the initial value of the feedback factor. Error threshold Feedback adjustment gradient Based on the parameter settings, network traffic prediction was performed on the simulated traffic sequence, and the traffic prediction results are as follows: Figure 7 As shown. From Figure 7 As can be seen, the predicted flow rate curve basically coincides with the actual flow rate curve. The correlation coefficient between the predicted and actual flow rate curves was calculated. The value approaching 1 indicates that the traffic prediction data of this invention is very close to the actual data, and the prediction effect is good. To verify the adaptive feedback gradient adjustment mechanism, during the traffic prediction process... The adjustment process was tracked, and the data tracking curve is as follows: Figure 8 As shown. Figure 8 middle, The value fluctuates, indicating that the adaptive feedback gradient adjustment mechanism is involved in the entire flow prediction.
[0128] To verify the universality of this method, we selected... and Network traffic is predicted separately, and the normalized absolute error of the two predictions is compared with... Compare the predicted normalized absolute value errors, such as... Figure 9As shown. Figure 9 It can be seen that in the initial stage of traffic prediction, and The normalized absolute error of the first three is relatively large. As the prediction progresses, the normalized absolute errors of the three gradually converge, indicating that... In the traffic prediction process, adaptive adjustment of the accuracy of network traffic prediction can be achieved, therefore The value of does not affect the accuracy of flow prediction, therefore the method of this invention has universality.
[0129] To verify the advantages of the method of the present invention, the traditional GM(1,1) grey prediction model and the ewboGM(1,1) grey prediction model with an added exponential weakened buffer operator were used to predict the simulated network traffic, and the prediction results were compared with the network traffic prediction results of the method of the present invention. The prediction effect is as follows: Figure 10 , Figure 11 and Figure 12 As shown.
[0130] in, Figure 10 This is a comparison chart of predicted and actual values based on the traditional GM(1,1) model. It shows a significant prediction error at the extreme values of the flow curve. The accuracy of the flow prediction has been calculated. ; Figure 11 This is a comparison chart of predicted and actual values based on the ewboGM(1,1) model. It shows that the prediction error has been somewhat mitigated, but significant prediction errors still exist at the extreme values of the flow curve. The accuracy of the flow prediction has been calculated. The accuracy of predictions has improved. Figure 12 The graph shows a comparison of the normalized absolute errors of the traffic prediction values and the actual values of the three models. It can be seen that the GM(1,1) gray prediction model has the largest normalized absolute error; although the normalized absolute error of the ewboGM(1,1) model has decreased, it is still significant; compared with the first two models, the traffic prediction method of this invention has the smallest normalized absolute error.
[0131] Calculations show that the flow prediction accuracy of the method of this invention is... Compared to the GM(1,1) model prediction It improved by 6.49%, compared to the prediction of the ewboGM(1,1) model. It improved by 1.55%, with the smallest normalized absolute error in traffic prediction. Under the conditions of Intel(R) Core(TM) i7-9700 CPU@3GHz processor, 32GB memory, and 64-bit Win10 operating system, the average prediction time of the algorithm is about 0.65 seconds.
[0132] While the present invention has been described above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.
Claims
1. A gray prediction method for network traffic based on an adaptive feedback adjustment mechanism, characterized in that, Based on the established network topology and the continuous traffic sequence to be predicted for that network topology. Perform the following steps S1 to S8 to obtain the target continuous flow sequence to be predicted. Predicted flow sequences; Step S1: Process the flow sequence Normalization is performed to obtain a normalized flow sequence. ; Step S2: Construct a model based on the correction factor Feedback correction function ; Step S3: Initialize the feedback correction function Feedback factor ,set up The initial value is ; Step S4: Based on the normalized flow sequence Feedback correction function and feedback correction function Feedback factor For normalized flow sequences A feedback correction function is applied to each element in the process. Obtain the preprocessed flow sequence ; Step S5: Use a sliding window prediction method to... Implement GM(1,1) grey prediction and retain the windowed prediction values at each time step in sequence; Step S6: Based on the window sliding prediction value at each time step S5, assign the window sliding prediction value at each time step to its corresponding value in... The flow rate values at the next time step are subtracted and compared to obtain the error of each flow rate sequence. Then, the error is predicted based on a preset threshold value. The relationship between the three predicted values and the actual values is obtained, including determining whether the flow sequence error is no greater than [value missing]. If yes, it is determined that the difference between the predicted value and the actual value is approximately equal to the preset range; otherwise, it is further determined whether the flow sequence error is greater than 0. If yes, it is determined that the predicted value is greater than the actual value; otherwise, it is determined that the predicted value is less than the actual value. Step S7: When the difference between the predicted value and the true value is approximately equal to a preset range, maintain the feedback factor for the window sliding prediction at the next time step. The size remains unchanged, that is Then return to step S4; When the predicted value is greater than the actual value, based on the preset feedback factor Corrected gradient Update the correction factor size for the next time-step window sliding prediction to... Then return to step S4; When the predicted value is less than the actual value, based on the preset feedback factor Corrected gradient Update the correction factor size for the next time-step window sliding prediction to... Then return to step S4; Step S8: Inverse normalize the series of flow prediction results obtained by traversing the flow sequence through the sliding window to obtain the predicted flow sequence.
2. The network traffic grey prediction method based on adaptive feedback adjustment mechanism according to claim 1, characterized in that, In step S2, the correction factor is constructed according to the following formula. Feedback correction function : ； in, It is a feedback factor.
3. The network traffic grey prediction method based on an adaptive feedback adjustment mechanism according to claim 2, characterized in that, Step S4 specifically involves generating the preprocessed flow sequence according to the following formula. : 。 4. The network traffic grey prediction method based on an adaptive feedback adjustment mechanism according to claim 3, characterized in that, Step S5 includes the following sub-steps: S5.1, For the flow sequence Initialize and obtain the initialized flow sequence. ; S5.2, The preset sliding prediction window size is m, and then... The m consecutively adjacent flow samples are used as the initialization sequence for each prediction. : ,in ; S5.3 Calculation A single accumulation generates a sequence : ； in, ； S5.4 Calculate according to the following formula The nearest neighbor mean generation sequence : ； in, ； S5.5, Construct according to the following formula and The grey differential equation: ； in, The gray development coefficient, This is the gray action quantity; S5.6 Construct the parameter vector of the grey differential equation from step S5.
5. And solve it using the least squares method according to the following formula: ， in, ， ； S5.7, obtained from the solution in step S5.6 , Calculate the time response function of the grey differential equation using the following formula. : ； S5.8, obtained from the solution in step S5.7 The prediction results are calculated using the following formula. : 。 5. The network traffic grey prediction method based on an adaptive feedback adjustment mechanism according to claim 4, characterized in that, The specific relationships between the three predicted values and the true values obtained in step S6 are as follows: S6.1 Judgment If the condition is met, then the difference between the predicted value and the actual value is approximately equal to the preset range. Otherwise, proceed to step S6.2; S6.2, Judgment If the condition is true, then the predicted value is greater than the actual value; otherwise, the predicted value is less than the actual value.
6. The network traffic grey prediction method based on an adaptive feedback adjustment mechanism according to claim 5, characterized in that, The network topology includes at least one traffic sending node and one traffic receiving node, with a P2P communication connection between the traffic sending node and the traffic receiving node.
7. The network traffic grey prediction method based on an adaptive feedback adjustment mechanism according to claim 6, characterized in that, The Wireshark packet capture software was used to continuously capture the traffic to be predicted in the network topology.