Machine learning based ipv6 traffic prediction system

Abstract

The application relates to the technical field of flow prediction, and discloses an IPv6 flow prediction system based on machine learning, which comprises a flow collection module, a multi-dimensional feature extraction module and a real-time prediction module.The flow collection module is used for capturing original flow data of an IPv6 network interface and performing protocol analysis and flow classification preprocessing to obtain preprocessed IPv6 flow data; the multi-dimensional feature extraction module is used for performing multi-modal feature extraction on the preprocessed IPv6 flow data to generate a fusion feature vector; the real-time prediction module is used for inputting the fusion feature vector into a dynamic prediction model to obtain an IPv6 flow prediction value in a future preset time length; and the dynamic scheduling module is used for pushing the IPv6 flow prediction value in the future preset time length to a flow scheduling controller to realize dynamic scheduling of network resources.The application improves the accuracy and real-time performance of network flow prediction, enhances the adaptability to IPv6 flow mode changes, and effectively helps to reasonably allocate and efficiently utilize network resources.

Description

Technical Field

[0001] This invention relates to the field of traffic prediction technology, and in particular to an IPv6 traffic prediction system based on machine learning. Background Technology

[0002] In today's digital age, accurate network traffic prediction is crucial for the rational allocation and efficient utilization of network resources. IPv6 (Internet Protocol Version 6), as the sixth version of the Internet Protocol, exhibits complex and variable traffic characteristics. Traditional traffic prediction methods struggle to meet the demands of accurate prediction in an IPv6 environment. Existing network traffic prediction technologies often suffer from insufficient prediction accuracy, poor real-time performance, and weak adaptability to changes in traffic patterns when handling large-scale, high-speed IPv6 data streams.

[0003] Therefore, there is an urgent need to provide a technical solution to address the above problems. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides an IPv6 traffic prediction system based on machine learning.

[0005] In a first aspect, the present invention provides an IPv6 traffic prediction system based on machine learning, the technical solution of which is as follows:

[0006] The traffic acquisition module is used to capture raw traffic data of the IPv6 network interface in real time through distributed probe nodes, and to perform protocol parsing and traffic classification preprocessing on the raw traffic data to obtain preprocessed IPv6 traffic data.

[0007] A multi-dimensional feature extraction module is used to extract multi-modal features from the preprocessed IPv6 traffic data and generate a fused feature vector.

[0008] The real-time prediction module is used to input the fused feature vector into the dynamic prediction model to obtain the IPv6 traffic prediction value for a preset time period in the future; the expression of the dynamic prediction model is:

[0009] ;in, Indicates the preset duration in the future. This represents the predicted IPv6 traffic value for a predetermined future time period. The dynamic prediction model is defined as follows: a machine learning model based on a lightweight inference engine, combining gated recurrent units and bidirectional recurrent neural networks. This represents a multidimensional feature extraction function; This refers to the preprocessed IPv6 traffic data;

[0010] The dynamic scheduling module is used to push the predicted IPv6 traffic for a preset future time period to the traffic scheduling controller to achieve dynamic scheduling of network resources.

[0011] The beneficial effects of the machine learning-based IPv6 traffic prediction system of the present invention are as follows:

[0012] The system of this invention can solve the problems of low prediction accuracy, poor real-time performance, and weak adaptability of traditional methods in IPv6 environment, improve the accuracy and real-time performance of network traffic prediction, enhance the adaptability to changes in IPv6 traffic patterns, and effectively help the rational allocation and efficient utilization of network resources.

[0013] Based on the above scheme, the IPv6 traffic prediction system based on machine learning of the present invention can be further improved as follows.

[0014] Furthermore, the raw traffic data includes: an IPv6 basic header and at least one extended header; the traffic acquisition module is specifically used for:

[0015] The IPv6 basic header is parsed to obtain the flow label field, and each extension header is parsed to obtain the security encapsulation header type and routing extension header information;

[0016] Based on the flow label field, the security encapsulation header type, and the routing extension header information, extract the protocol type identifier and routing path parameters;

[0017] Based on the protocol type identifier, IPv6 packets belonging to the same service type in the original traffic data are aggregated into service type subsets, and each service type subset is processed by packet length normalization and timestamp alignment to generate the preprocessed IPv6 traffic data.

[0018] Furthermore, the multidimensional feature extraction module is specifically used for:

[0019] The preprocessed IPv6 traffic data is extracted using a sliding time window to obtain a statistical feature vector;

[0020] Based on the protocol type identifier and routing path parameters, the preprocessed IPv6 traffic data is processed to generate a protocol stack feature vector.

[0021] The preprocessed IPv6 traffic data is extracted using discrete wavelet transform to obtain multi-scale time-series feature vectors;

[0022] After normalizing the statistical feature vector, the protocol stack feature vector, and the multi-scale temporal feature vector, the fused feature vector is generated by vector concatenation; the expression of the fused feature vector is:

[0023] ;in, Indicates the size of the sliding time window. This represents the flow rate value in the i-th time window. Represents the statistical feature vector, This represents the feature vector of the protocol stack. This represents the one-hot encoding function. This represents the multi-scale temporal feature vector. Represents discrete wavelet transform. This indicates the stream label field. This represents the cross product operation of vectors. This represents the routing path parameters.

[0024] Furthermore, the dynamic scheduling module is specifically used for:

[0025] The predicted IPv6 traffic for a preset future duration is pushed to the traffic scheduling controller, so that the traffic scheduling controller can generate a bandwidth allocation instruction based on the predicted IPv6 traffic for a preset future duration to achieve dynamic scheduling of network resources; the expression of the bandwidth allocation instruction is:

[0026] ;in, This represents the bandwidth allocation value for the preset future duration. This indicates the current bandwidth utilization rate. This represents the weighting coefficient of the predicted value.

[0027] Furthermore, it also includes:

[0028] An online incremental learning module is used to dynamically update the parameters of the dynamic prediction model based on a time decay weighting mechanism. The time decay weighting mechanism allocates training weights according to the timestamps of the preprocessed IPv6 traffic data, enabling the dynamic prediction model to adapt to IPv6 traffic feature drift. The expression for dynamically updating the parameters of the dynamic prediction model is:

[0029] , ;in, This represents the parameters of the dynamic prediction model at time t. This represents the learning rate of the dynamic prediction model. Represents the loss function. express Time decay weight, Indicates the attenuation coefficient. Indicates sample timestamp, Indicates the current timestamp. The first part of the dynamic prediction model represents the... The actual traffic value of each sample The first part of the dynamic prediction model represents the... The predicted flow value for each sample.

[0030] Furthermore, the dynamic scheduling module is also used for:

[0031] Before generating the bandwidth allocation instruction, the following anomaly detection logic is executed:

[0032] when hour, ;in, This represents the historical average traffic volume. Indicates the standard deviation of historical flow. Indicates the abnormal threshold. This represents the capacity expansion coefficient for sudden traffic surges. .

[0033] Furthermore, the attenuation coefficient of the time-attenuation weighted mechanism is dynamically adjusted according to the flow rate change, and its expression is:

[0034] ;in, Indicates the basic attenuation coefficient. This represents the first derivative of the traffic forecast value with respect to time. This represents the hyperbolic tangent function.

[0035] Secondly, this invention provides an IPv6 traffic prediction method based on machine learning, the technical solution of which is as follows:

[0036] The raw traffic data of the IPv6 network interface is captured in real time by distributed probe nodes, and the raw traffic data is preprocessed by protocol parsing and traffic classification to obtain preprocessed IPv6 traffic data.

[0037] Multimodal feature extraction is performed on the preprocessed IPv6 traffic data to generate a fused feature vector;

[0038] The fused feature vector is input into the dynamic prediction model to obtain the IPv6 traffic prediction value for a preset time period in the future; the expression of the dynamic prediction model is:

[0039] ;in, Indicates the preset duration in the future. This represents the predicted IPv6 traffic value for a predetermined future time period. The dynamic prediction model is defined as follows: a machine learning model based on a lightweight inference engine, combining gated recurrent units and bidirectional recurrent neural networks. This represents a multidimensional feature extraction function; This refers to the preprocessed IPv6 traffic data;

[0040] The predicted IPv6 traffic for a preset future duration is pushed to the traffic scheduling controller to achieve dynamic scheduling of network resources.

[0041] The beneficial effects of the machine learning-based IPv6 traffic prediction method of the present invention are as follows:

[0042] The method of this invention can solve the problems of low prediction accuracy, poor real-time performance, and weak adaptability of traditional methods in IPv6 environment, improve the accuracy and real-time performance of network traffic prediction, enhance the adaptability to changes in IPv6 traffic patterns, and effectively help the rational allocation and efficient utilization of network resources.

[0043] Based on the above scheme, the IPv6 traffic prediction method based on machine learning of the present invention can be further improved as follows.

[0044] Furthermore, the raw traffic data includes: an IPv6 basic header and at least one extended header; the step of performing protocol parsing and traffic classification preprocessing on the raw traffic data to obtain preprocessed IPv6 traffic data further includes:

[0045] The IPv6 basic header is parsed to obtain the flow label field, and each extension header is parsed to obtain the security encapsulation header type and routing extension header information;

[0046] Based on the flow label field, the security encapsulation header type, and the routing extension header information, extract the protocol type identifier and routing path parameters;

[0047] Based on the protocol type identifier, IPv6 packets belonging to the same service type in the original traffic data are aggregated into service type subsets, and each service type subset is processed by packet length normalization and timestamp alignment to generate the preprocessed IPv6 traffic data.

[0048] Furthermore, the step of extracting multimodal features from the preprocessed IPv6 traffic data and generating a fused feature vector further includes:

[0049] The preprocessed IPv6 traffic data is extracted using a sliding time window to obtain a statistical feature vector;

[0050] Based on the protocol type identifier and routing path parameters, the preprocessed IPv6 traffic data is processed to generate a protocol stack feature vector.

[0051] The preprocessed IPv6 traffic data is extracted using discrete wavelet transform to obtain multi-scale time-series feature vectors;

[0052] After normalizing the statistical feature vector, the protocol stack feature vector, and the multi-scale temporal feature vector, the fused feature vector is generated by vector concatenation; the expression of the fused feature vector is:

[0053] ;in, Indicates the size of the sliding time window. This represents the flow rate value in the i-th time window. Represents the statistical feature vector, This represents the feature vector of the protocol stack. This represents the one-hot encoding function. This represents the multi-scale temporal feature vector. Represents discrete wavelet transform. This indicates the stream label field. This represents the cross product operation of vectors. This represents the routing path parameters.

[0054] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0057] Figure 1 This is a schematic diagram of an embodiment of the IPv6 traffic prediction system based on machine learning according to the present invention.

[0058] Figure 2This is a flowchart illustrating an embodiment of an IPv6 traffic prediction method based on machine learning according to the present invention. Detailed Implementation

[0059] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0060] Figure 1 A schematic diagram of an embodiment of an IPv6 traffic prediction system based on machine learning provided by the present invention is shown. Figure 1 As shown, the system includes: a traffic acquisition module 110, a multi-dimensional feature extraction module 120, a real-time prediction module 130, and a dynamic scheduling module 140;

[0061] The traffic acquisition module 110 is used to capture raw traffic data of the IPv6 network interface in real time through distributed probe nodes, and to perform protocol parsing and traffic classification preprocessing on the raw traffic data to obtain preprocessed IPv6 traffic data.

[0062] The multi-dimensional feature extraction module 120 is used to extract multi-modal features from the preprocessed IPv6 traffic data and generate a fused feature vector.

[0063] The real-time prediction module 130 is used to input the fused feature vector into the dynamic prediction model to obtain the IPv6 traffic prediction value for a preset time period in the future; the expression of the dynamic prediction model is:

[0064] ;in, Indicates the preset duration in the future. This represents the predicted IPv6 traffic value for a predetermined future time period. The dynamic prediction model is defined as follows: a machine learning model based on a lightweight inference engine, combining gated recurrent units and bidirectional recurrent neural networks. This represents a multidimensional feature extraction function; This refers to the preprocessed IPv6 traffic data;

[0065] The dynamic scheduling module 140 is used to push the predicted IPv6 traffic value for a preset future duration to the traffic scheduling controller to realize dynamic scheduling of network resources.

[0066] It should be noted that distributed probe nodes refer to data collection agents deployed on critical link nodes of the IPv6 network. They capture raw data packets flowing through network interfaces in real time through port mirroring or traffic replication technologies. The nodes work collaboratively in a distributed architecture to ensure comprehensive collection of high-speed IPv6 traffic and achieve load balancing through a consistent hashing algorithm. IPv6 network interfaces refer to the physical or virtual interfaces of network devices that support the IPv6 protocol stack, used to send and receive IPv6 packets conforming to the RFC 2460 standard. IPv6 network interfaces include 10 Gigabit Ethernet ports on routers, VXLAN tunnel endpoints on virtual switches, etc., and their traffic carrying capacity must meet line-speed processing requirements of 100Gbps or higher. Raw traffic data refers to the set of unprocessed IPv6 data packets captured by the distributed probe nodes, including the IPv6 basic header (source / destination address, flow label, hop count limit) and at least one extended header (such as a routing extension header or a security encapsulation header). Raw traffic data is stored in PCAP format, retaining complete protocol layer information and timestamp metadata.

[0067] In one alternative embodiment, the traffic acquisition module 110 is specifically used for:

[0068] The IPv6 basic header is parsed to obtain the flow label field, and each extension header is parsed to obtain the security encapsulation header type and routing extension header information.

[0069] Based on the flow label field, the security encapsulation header type, and the routing extension header information, extract the protocol type identifier and routing path parameters.

[0070] Based on the protocol type identifier, IPv6 packets belonging to the same service type in the original traffic data are aggregated into service type subsets, and each service type subset is processed by packet length normalization and timestamp alignment to generate the preprocessed IPv6 traffic data.

[0071] It's important to note that the Flow Label field is a 20-bit field in the IPv6 basic header used to identify data packets within the same service flow, ensuring differentiated processing by network devices for specific service types (such as video streams and real-time communication). The Security Encapsulation Header Type is a field in the IPv6 extended header that identifies the data encryption and authentication protocol type, including ESP (Encapsulating Security Payload, IPsec encryption) and AH (Authentication Header, IPsec Integrity Check). The Routing Extension Header information refers to the cross-domain transmission path parameters recorded in the IPv6 Routing Extension Header (Type 0, RFC 8200), including a list of intermediate nodes, remaining hop count, and routing type identifier. The Protocol Type Identifier is the service type code generated by parsing the Flow Label field and the Security Encapsulation Header, used to distinguish traffic from different application scenarios (such as video streams, encrypted control signaling, and IoT sensor data). Routing path parameters are quantitative indicators extracted from the routing extension header information, used to describe traffic transmission path characteristics, including cross-domain hop count, path node change frequency, and routing type (strict / loose routing). A service type subset refers to a set of IPv6 packets aggregated according to the protocol type identifier. Traffic within the same subset has the same service attributes and traffic patterns (such as a video stream subset containing long packets and periodic transmission characteristics).

[0072] In an alternative embodiment, the multidimensional feature extraction module 120 is specifically used for:

[0073] The preprocessed IPv6 traffic data is extracted using a sliding time window to obtain a statistical feature vector;

[0074] Based on the protocol type identifier and routing path parameters, the preprocessed IPv6 traffic data is processed to generate a protocol stack feature vector.

[0075] The preprocessed IPv6 traffic data is extracted using discrete wavelet transform to obtain multi-scale time-series feature vectors;

[0076] After normalizing the statistical feature vector, the protocol stack feature vector, and the multi-scale temporal feature vector, the fused feature vector is generated by vector concatenation; the expression of the fused feature vector is:

[0077] ;in, Indicates the size of the sliding time window. This represents the flow rate value in the i-th time window. Represents the statistical feature vector, This represents the feature vector of the protocol stack. This represents the one-hot encoding function. This represents the multi-scale temporal feature vector. Represents discrete wavelet transform. This indicates the stream label field. This represents the cross product operation of vectors. This represents the routing path parameters.

[0078] In an alternative embodiment, the dynamic scheduling module 140 is specifically used for:

[0079] The predicted IPv6 traffic for a preset future duration is pushed to the traffic scheduling controller, so that the traffic scheduling controller can generate a bandwidth allocation instruction based on the predicted IPv6 traffic for a preset future duration to achieve dynamic scheduling of network resources; the expression of the bandwidth allocation instruction is:

[0080] ;in, This represents the bandwidth allocation value for the preset future duration. This indicates the current bandwidth utilization rate. This represents the weighting coefficient of the predicted value.

[0081] In one alternative approach, it also includes:

[0082] An online incremental learning module is used to dynamically update the parameters of the dynamic prediction model based on a time decay weighting mechanism. The time decay weighting mechanism allocates training weights according to the timestamps of the preprocessed IPv6 traffic data, enabling the dynamic prediction model to adapt to IPv6 traffic feature drift. The expression for dynamically updating the parameters of the dynamic prediction model is:

[0083] , ;in, This represents the parameters of the dynamic prediction model at time t. This represents the learning rate of the dynamic prediction model. Represents the loss function. express Time decay weight, Indicates the attenuation coefficient. Indicates sample timestamp, Indicates the current timestamp. The first part of the dynamic prediction model represents the... The actual traffic value of each sample The first part of the dynamic prediction model represents the... The predicted flow value for each sample.

[0084] In an alternative embodiment, the dynamic scheduling module 140 is further configured to:

[0085] Before generating the bandwidth allocation instruction, the following anomaly detection logic is executed:

[0086] when hour, ;in, This represents the historical average traffic volume. Indicates the standard deviation of historical flow. Indicates the abnormal threshold. This represents the capacity expansion coefficient for sudden traffic surges. .

[0087] In one alternative approach, the attenuation coefficient of the time-attenuation weighted mechanism is dynamically adjusted according to the flow rate change, expressed as:

[0088] ;in, Indicates the basic attenuation coefficient. This represents the first derivative of the traffic forecast value with respect to time. This represents the hyperbolic tangent function.

[0089] The technical solution of this embodiment can solve the problems of low prediction accuracy, poor real-time performance, and weak adaptability of traditional methods in the IPv6 environment. It improves the accuracy and real-time performance of network traffic prediction, enhances the adaptability to changes in IPv6 traffic patterns, and effectively helps to rationally allocate and efficiently utilize network resources.

[0090] Figure 2 This diagram illustrates a flowchart of an embodiment of a machine learning-based IPv6 traffic prediction method provided by the present invention. Figure 2 As shown, it includes the following steps:

[0091] S1. The raw traffic data of the IPv6 network interface is captured in real time through distributed probe nodes, and the raw traffic data is preprocessed by protocol parsing and traffic classification to obtain preprocessed IPv6 traffic data.

[0092] S2. Perform multimodal feature extraction on the preprocessed IPv6 traffic data to generate a fused feature vector;

[0093] S3. Input the fused feature vector into the dynamic prediction model to obtain the IPv6 traffic prediction value for a preset time period in the future; the expression of the dynamic prediction model is:

[0094] ;in, Indicates the preset duration in the future. This represents the predicted IPv6 traffic value for a predetermined future time period. The dynamic prediction model is defined as follows: a machine learning model based on a lightweight inference engine, combining gated recurrent units and bidirectional recurrent neural networks. This represents a multidimensional feature extraction function; This refers to the preprocessed IPv6 traffic data;

[0095] S4. Push the predicted IPv6 traffic value for the preset future duration to the traffic scheduling controller to achieve dynamic scheduling of network resources.

[0096] In one optional approach, the raw traffic data includes: an IPv6 basic header and at least one extended header; the step of performing protocol parsing and traffic classification preprocessing on the raw traffic data to obtain preprocessed IPv6 traffic data further includes:

[0097] The IPv6 basic header is parsed to obtain the flow label field, and each extension header is parsed to obtain the security encapsulation header type and routing extension header information;

[0098] Based on the flow label field, the security encapsulation header type, and the routing extension header information, extract the protocol type identifier and routing path parameters;

[0099] Based on the protocol type identifier, IPv6 packets belonging to the same service type in the original traffic data are aggregated into service type subsets, and each service type subset is processed by packet length normalization and timestamp alignment to generate the preprocessed IPv6 traffic data.

[0100] In one alternative approach, the step of extracting multimodal features from the preprocessed IPv6 traffic data to generate a fused feature vector further includes:

[0101] The preprocessed IPv6 traffic data is extracted using a sliding time window to obtain a statistical feature vector;

[0102] Based on the protocol type identifier and routing path parameters, the preprocessed IPv6 traffic data is processed to generate a protocol stack feature vector.

[0103] The preprocessed IPv6 traffic data is extracted using discrete wavelet transform to obtain multi-scale time-series feature vectors;

[0104] After normalizing the statistical feature vector, the protocol stack feature vector, and the multi-scale temporal feature vector, the fused feature vector is generated by vector concatenation; the expression of the fused feature vector is:

[0105] ;in, Indicates the size of the sliding time window. This represents the flow rate value in the i-th time window. Represents the statistical feature vector, This represents the feature vector of the protocol stack. This represents the one-hot encoding function. This represents the multi-scale temporal feature vector. Represents discrete wavelet transform. This indicates the stream label field. This represents the cross product operation of vectors. This represents the routing path parameters.

[0106] The technical solution of this embodiment can solve the problems of low prediction accuracy, poor real-time performance, and weak adaptability of traditional methods in the IPv6 environment. It improves the accuracy and real-time performance of network traffic prediction, enhances the adaptability to changes in IPv6 traffic patterns, and effectively helps to rationally allocate and efficiently utilize network resources.

[0107] Furthermore, the system provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the system can be divided into different functional modules according to the actual situation to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0108] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0109] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0110] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1.A machine learning based IPv6 traffic prediction system, characterized in that, The system comprises: a traffic collection module for capturing raw traffic data of an IPv6 network interface in real time through a distributed probe node, and performing protocol analysis and traffic classification preprocessing on the raw traffic data to obtain preprocessed IPv6 traffic data; a multi-dimensional feature extraction module for performing multi-modal feature extraction on the preprocessed IPv6 traffic data to generate a fusion feature vector; a real-time prediction module for inputting the fusion feature vector into a dynamic prediction model to obtain an IPv6 traffic prediction value for a future preset time length; the expression of the dynamic prediction model is: wherein, represents a future preset time length, represents an IPv6 traffic prediction value of the future preset time length, represents the dynamic prediction model; the dynamic prediction model is a machine learning model deployed based on a lightweight inference engine, combining a gated recurrent unit and a bidirectional recurrent neural network; represents a multi-dimensional feature extraction function; represents the preprocessed IPv6 traffic data; a dynamic scheduling module for pushing the IPv6 traffic prediction value for the future preset time length to a traffic scheduling controller to realize dynamic scheduling of network resources; the multi-dimensional feature extraction module is specifically configured to: extract the preprocessed IPv6 traffic data through a sliding time window to obtain a statistical feature vector; process the preprocessed IPv6 traffic data based on protocol type identification and routing path parameters to generate a protocol stack feature vector; extract the preprocessed IPv6 traffic data using discrete wavelet transform to obtain a multi-scale time series feature vector; normalize the statistical feature vector, the protocol stack feature vector, and the multi-scale time series feature vector, and then generate the fusion feature vector through vector splicing; the expression of the fusion feature vector is: wherein, denotes a sliding time window size, denotes a traffic value of the i-th time window, denotes the statistical feature vector, denotes the protocol stack feature vector, denotes a one-hot encoding function, denotes the multi-scale time series feature vector, denotes a discrete wavelet transform, denotes a flow label field, denotes a vector outer product operation, denotes the routing path parameter; further comprising: an online incremental learning module for dynamically updating parameters of the dynamic prediction model based on a time decay weighting mechanism; wherein the time decay weighting mechanism assigns a training weight according to a timestamp of the preprocessed IPv6 traffic data, so that the dynamic prediction model adapts to IPv6 traffic feature drift; the expression of dynamically updating the parameters of the dynamic prediction model is: ， wherein, denotes a parameter of the dynamic prediction model at time t, denotes a learning rate of the dynamic prediction model, denotes a loss function, denotes a time decay weight of denotes a decay coefficient, denotes a sample a timestamp of denotes a current timestamp, denotes a true traffic value of the dynamic prediction model for a denotes a predicted traffic value of the dynamic prediction model for a denotes a predicted traffic value of the dynamic prediction model for a denotes a predicted traffic value of the dynamic prediction model for a 2.The machine learning based IPv6 traffic prediction system of claim 1, wherein, the raw traffic data includes an IPv6 basic header and at least one extension header; the traffic collection module is specifically configured to: analyze the IPv6 basic header to obtain a flow label field, and analyze each extension header to obtain security encapsulation header types and routing extension header information; extract protocol type identification and routing path parameters according to the flow label field, the security encapsulation header types, and the routing extension header information; according to the protocol type identification, aggregate IPv6 data packets belonging to the same service type in the raw traffic data into a service type subset, and perform packet length normalization and timestamp alignment processing on each service type subset to generate the preprocessed IPv6 traffic data. 3.The machine learning based IPv6 traffic prediction system of claim 1, wherein, the dynamic scheduling module is specifically configured to: push the IPv6 traffic prediction value for the future preset time length to the traffic scheduling controller, so that the traffic scheduling controller generates a bandwidth allocation instruction according to the IPv6 traffic prediction value for the future preset time length to realize dynamic scheduling of network resources; the expression of the bandwidth allocation instruction is: wherein, represents a bandwidth allocation value of the future preset time length, represents a current bandwidth utilization rate, represents a prediction value weight coefficient. 4.The machine learning based IPv6 traffic prediction system of claim 3, wherein, the dynamic scheduling module is further configured to: before generating the bandwidth allocation instruction, execute the following exception detection logic: When , ; wherein, represents the historical traffic mean value, represents the historical traffic standard deviation, represents the abnormal threshold value, represents the burst traffic expansion coefficient, . 5.The machine learning based IPv6 traffic prediction system of claim 1, wherein, the decay coefficient of the time decay weighting mechanism is dynamically adjusted according to the traffic change rate, and the expression is: wherein denotes the base attenuation coefficient, denotes the first derivative of the flow prediction value with respect to time, denotes the hyperbolic tangent function. 6.A method for IPv6 traffic prediction based on machine learning, characterized in that, The method comprises: Real-time capture of raw traffic data of an IPv6 network interface by a distributed probe node, and protocol analysis and traffic classification preprocessing of the raw traffic data to obtain preprocessed IPv6 traffic data; Multi-modal feature extraction of the preprocessed IPv6 traffic data to generate a fusion feature vector; Inputting the fusion feature vector into a dynamic prediction model to obtain an IPv6 traffic prediction value for a future preset time length; the expression of the dynamic prediction model is: wherein, represents a future preset time length, represents an IPv6 traffic prediction value of the future preset time length, represents the dynamic prediction model; the dynamic prediction model is a machine learning model deployed based on a lightweight inference engine, combining a gated recurrent unit and a bidirectional recurrent neural network; represents a multi-dimensional feature extraction function; represents the preprocessed IPv6 traffic data; Pushing the IPv6 traffic prediction value for the future preset time length to a traffic scheduling controller to realize dynamic scheduling of network resources; The step of multi-modal feature extraction of the preprocessed IPv6 traffic data to generate a fusion feature vector further comprises: Extraction of the preprocessed IPv6 traffic data by a sliding time window to obtain a statistical feature vector; Processing of the preprocessed IPv6 traffic data based on protocol type identification and routing path parameters to generate a protocol stack feature vector; Extraction of the preprocessed IPv6 traffic data using discrete wavelet transform to obtain a multi-scale time series feature vector; After normalization processing of the statistical feature vector, the protocol stack feature vector and the multi-scale time series feature vector, the fusion feature vector is generated by vector splicing; the expression of the fusion feature vector is: wherein, denotes a sliding time window size, denotes a traffic value of the i-th time window, denotes the statistical feature vector, denotes the protocol stack feature vector, denotes a one-hot encoding function, denotes the multi-scale time series feature vector, denotes a discrete wavelet transform, denotes a flow label field, denotes a vector outer product operation, denotes the routing path parameter; The method further comprises: Dynamic updating of parameters of the dynamic prediction model based on a time decay weighting mechanism; wherein the time decay weighting mechanism assigns a training weight according to a timestamp of the preprocessed IPv6 traffic data, so that the dynamic prediction model is self-adaptive to IPv6 traffic feature drift; the expression of dynamic updating of parameters of the dynamic prediction model is: ， wherein, denotes a parameter of the dynamic prediction model at time t, denotes a learning rate of the dynamic prediction model, denotes a loss function, denotes a time decay weight of the dynamic prediction model, denotes a decay coefficient, denotes a sample a timestamp of the sample, denotes a current timestamp, denotes a true traffic value of the dynamic prediction model for a th sample, denotes a predicted traffic value of the dynamic prediction model for a th sample. 7.The method of claim 6, wherein, The raw traffic data comprises an IPv6 basic header and at least one extension header; the step of protocol analysis and traffic classification preprocessing of the raw traffic data to obtain preprocessed IPv6 traffic data further comprises: Analysis of the IPv6 basic header to obtain a flow label field, and analysis of each extension header to obtain security encapsulation header types and routing extension header information; Extraction of protocol type identification and routing path parameters according to the flow label field, the security encapsulation header types and the routing extension header information; According to the protocol type identification, aggregation of IPv6 data packets belonging to the same service type in the raw traffic data into a service type subset, and packet length normalization and timestamp alignment processing of each service type subset to generate the preprocessed IPv6 traffic data.

Images