Traffic prediction and dynamic scheduling method for industrial TSN network

Abstract

The embodiment of the invention provides an industrial TSN network-oriented traffic prediction and dynamic scheduling method, which is applied to the technical field of network traffic management. The method comprises the following steps: collecting data from an industrial TSN network and pre-processing the data to generate a data packet sequence sorted based on a time sequence; based on task requirements and network characteristics, extracting flow characteristics from the data packet sequence to construct a data fitting model; fitting the data packet sequence based on a data fitting model to generate fitting time sequence data; performing clustering analysis on the fitting time sequence data, predicting a future network flow change trend, and obtaining a corresponding preset strategy according to a prediction result; and based on the prediction result and a corresponding preset strategy, dynamically adjusting network configuration, and carrying out traffic shaping and priority scheduling. In this way, the problem that in the prior art, complex and changeable flow requirements in an industrial TSN network environment are difficult to meet can be solved, and efficient utilization of network resources and optimization of network performance are achieved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of network traffic management, and in particular to a traffic prediction and dynamic scheduling method for industrial TSN networks. Background Art

[0002] In industrial Time-Sensitive Networking (TSN), network traffic prediction and optimization face a series of unique technical challenges. Compared with ordinary Ethernet, TSN networks must not only meet the basic requirements of high bandwidth and low latency, but also ensure the real-time transmission and precise synchronization of time-sensitive data, which is crucial for ensuring critical communications in industrial control and aerospace missions.

[0003] However, network traffic in industrial applications is often highly bursty and periodic, and is affected by multiple external factors, such as device status and task scheduling. This makes traffic prediction extremely complex. In addition, network delay and jitter are difficult to avoid in practical applications, which also places higher demands on the real-time and accuracy of traffic prediction and optimization.

[0004] In response to the above technical problems, the present disclosure provides a traffic prediction and dynamic scheduling method for industrial TSN networks. Based on this method, network traffic can be captured and analyzed in real time, future traffic trends can be predicted through intelligent prediction algorithms, and refined management can be performed based on the prediction results and preset strategies to achieve efficient utilization of network resources and optimization of network performance, thereby enhancing network security and reliability. Summary of the Invention

[0005] The present disclosure provides a traffic prediction and dynamic scheduling method for industrial TSN networks, which solves the technical problem that most existing technologies rely on traditional traffic monitoring tools, lack intelligent and refined management capabilities, and are difficult to cope with the complex and changeable traffic demands in the TSN network environment.

[0006] According to a first aspect of the present disclosure, a method for traffic prediction and dynamic scheduling for industrial TSN networks is provided. The method comprises:

[0007] Collect and pre-process data from industrial TSN networks to generate data packet sequences based on time series sorting;

[0008] Based on task requirements and network characteristics, traffic features are extracted from the data packet sequence to build a data fitting model;

[0009] Fitting the data packet sequence based on the data fitting model to generate fitting time series data;

[0010] Performing cluster analysis on the fitted time series data to predict future network traffic change trends, and obtaining corresponding preset strategies based on the prediction results;

[0011] Based on the prediction results and corresponding preset strategies, the network configuration is dynamically adjusted to perform traffic shaping and priority scheduling.

[0012] The above aspects and any possible implementation manner further provide an implementation manner, wherein extracting traffic features from the data packet sequence based on task requirements and network characteristics to construct a data fitting model includes:

[0013] Extracting traffic features from the data packet sequence based on task requirements and network characteristics, and performing fusion calculation on each traffic feature to generate a traffic feature vector;

[0014] Based on the traffic characteristics, a statistical regression method is used to fit the traffic change trend, and a Fourier transform algorithm is used to detect the periodic characteristics of the traffic;

[0015] Based on the flow variation trend and the periodic characteristics of the flow, a probability distribution algorithm is used to model the flow feature vector to generate a data fitting model.

[0016] According to the above aspect and any possible implementation, an implementation is further provided, wherein fitting the data packet sequence based on the data fitting model to generate fitting time series data includes:

[0017] Fitting the mean data and the fluctuation data in the traffic feature vector based on the data fitting model to generate mean time series data and fluctuation time series data;

[0018] The mean time series data and the fluctuation time series data are synthesized according to the time series correspondence to generate fitting time series data.

[0019] According to the above aspects and any possible implementation, an implementation is further provided, wherein the traffic characteristics include length distribution, arrival time interval, burst rate and periodic characteristics of data packets.

[0020] According to the above aspects and any possible implementation, a further implementation is provided, wherein cluster analysis is performed on the fitted time series data to predict future network traffic change trends, and corresponding preset strategies are obtained based on the prediction results, including:

[0021] Performing cluster analysis on the fitted time series data based on a clustering algorithm to divide the data into different clusters, wherein each cluster represents a traffic pattern and / or traffic change trend;

[0022] Based on the time series prediction model, the traffic data in each cluster is analyzed separately, and the analysis results of each cluster are combined to predict the future traffic change trend;

[0023] Based on the prediction results, the corresponding preset strategy is obtained from the preset strategy library.

[0024] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein the method further includes:

[0025] Encrypt sensitive data and traffic information during data transmission and control user access rights to data.

[0026] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein the method further includes:

[0027] Conduct real-time detection of network traffic during data transmission, and take filtering and control measures for abnormal traffic.

[0028] According to the above aspects and any possible implementation manner, an implementation manner is further provided, wherein the method further includes:

[0029] Visualize the overall status of network traffic and traffic prediction results during data transmission.

[0030] According to a second aspect of the present disclosure, a traffic prediction and dynamic scheduling device for an industrial TSN network is provided. The device includes:

[0031] The data acquisition module is used to collect and pre-process data from the industrial TSN network to generate a sequence of data packets based on time series sorting;

[0032] A model building module is used to extract traffic features from the data packet sequence and build a data fitting model based on task requirements and network characteristics;

[0033] A data fitting module, configured to fit the data packet sequence based on the data fitting model to generate fitting time series data;

[0034] An analysis and prediction module is used to perform cluster analysis on the fitted time series data, predict future network traffic change trends, and obtain corresponding preset strategies based on the prediction results;

[0035] The adjustment and optimization module is used to dynamically adjust the network configuration based on the prediction results and the corresponding preset strategies, and perform traffic shaping and priority scheduling.

[0036] According to a third aspect of the present disclosure, an electronic device is provided, comprising: a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the above method when executing the program.

[0037] In the embodiment of the present disclosure, data is first collected from the industrial TSN network and preprocessed to generate a data packet sequence based on time series sorting; based on task requirements and network characteristics, traffic features are extracted from the data packet sequence to construct a data fitting model; secondly, based on the data fitting model, the data packet sequence is fitted to generate fitting time series data; cluster analysis is performed on the fitting time series data to predict future network traffic change trends, and corresponding preset strategies are obtained based on the prediction results; finally, based on the prediction results and the corresponding preset strategies, the network configuration is dynamically adjusted to perform traffic shaping and priority scheduling. In this way, the problem that most existing technologies rely on traditional traffic monitoring tools, lack intelligent and refined management capabilities, and are difficult to cope with the complex and changing traffic demands in the industrial TSN network environment can be solved, so as to achieve real-time capture and analysis of network traffic, predict future traffic trends through intelligent prediction algorithms, and perform refined management based on the prediction results and preset strategies, further achieving efficient utilization of network resources and optimization of network performance, making it better adapted to different network environments and business needs.

[0038] It should be understood that the contents described in the Summary of the Invention section are not intended to limit the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The above and other features, advantages, and aspects of the various embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. The accompanying drawings are provided for a better understanding of the present disclosure and do not constitute a limitation of the present disclosure. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, among which:

[0040] Figure 1 A flow chart of a method for traffic prediction and dynamic scheduling for an industrial TSN network provided by an embodiment of the present disclosure is shown;

[0041] Figure 2 A structural diagram of a traffic prediction and dynamic scheduling device for an industrial TSN network provided by an embodiment of the present disclosure is shown;

[0042] Figure 3 A structural diagram of an exemplary electronic device capable of implementing an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0043] To make the purpose, technical solutions, and advantages of the embodiments of the present disclosure more clear, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present disclosure.

[0044] In this document, the term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the related objects are in an "or" relationship.

[0045] In the embodiment of the present disclosure, data is first collected from the industrial TSN network and preprocessed to generate a data packet sequence based on time series sorting; based on task requirements and network characteristics, traffic features are extracted from the data packet sequence to construct a data fitting model; secondly, based on the data fitting model, the data packet sequence is fitted to generate fitting time series data; cluster analysis is performed on the fitting time series data to predict future network traffic change trends, and corresponding preset strategies are obtained based on the prediction results; finally, based on the prediction results and the corresponding preset strategies, the network configuration is dynamically adjusted to perform traffic shaping and priority scheduling. In this way, the problem that most existing technologies rely on traditional traffic monitoring tools, lack intelligent and refined management capabilities, and are difficult to cope with the complex and changing traffic demands in the industrial TSN network environment can be solved, so as to achieve real-time capture and analysis of network traffic, predict future traffic trends through intelligent prediction algorithms, and perform refined management based on the prediction results and preset strategies, further achieving efficient utilization of network resources and optimization of network performance, making it better adapted to different network environments and business needs.

[0046] This paper provides a traffic prediction and dynamic scheduling method for industrial TSN networks, which is applied to traffic model measurement and analysis equipment. The equipment is connected to the industrial TSN network interface via an FPGA. The purpose is to monitor and analyze network traffic in the industrial TSN network environment in real time, predict future traffic trends, and optimize network configuration accordingly to improve network performance and resource utilization. The core functions of the equipment are all implemented through a high-performance FPGA, ensuring the real-time and accuracy of the prediction.

[0047] The following describes in detail a traffic prediction and dynamic scheduling method for an industrial TSN network provided by an embodiment of the present disclosure through specific embodiments in conjunction with the accompanying drawings.

[0048] Figure 1FIG1 shows a flow chart of a method for traffic prediction and dynamic scheduling for an industrial TSN network provided by an embodiment of the present disclosure, as shown in FIG1. Figure 1 As shown, a traffic prediction and dynamic scheduling method 100 for an industrial TSN network may include the following steps:

[0049] S110 collects and preprocesses data from the industrial TSN network to generate a data packet sequence based on time series sorting.

[0050] For example, a high-performance FPGA (field programmable gate array) deployed on a network node is first connected to the industrial TSN network (time-sensitive network) interface to monitor and analyze data packets in the industrial TSN network in real time. The FPGA transfers the captured data packets to memory (such as DDR) through its built-in DMA (direct memory access) engine and uses a CRC (cyclic redundancy check) mechanism to check the integrity of the data packets during data transmission, eliminating any damaged or incomplete packets. In addition, the FPGA parses the data packets, extracts key fields (such as source / destination address, port information, protocol type, etc.), and classifies and stores them based on the data packet's protocol, source / destination address, timestamp, and other features to generate a time-series-ordered data packet sequence. It also normalizes the data packet feature values (such as rate and burst rate) for unified analysis.

[0051] Exemplarily, the specific steps of the data packet integrity verification include: generating and verifying a CRC (cyclic redundancy check) value for each data packet, and discarding the packet if the verification fails;

[0052] The FPGA ensures the integrity of the data frame by comparing the start and end characters of the data packet (only complete data packets will be stored and passed to the downstream processing module);

[0053] Captured data packets are temporarily stored based on FIFO (first-in-first-out) queues or BRAM (block RAM) to ensure smooth data flow. The parallel computing capability of FPGA can process multiple data packets simultaneously to avoid delays.

[0054] For example, captured data packets are parsed and key fields (such as source / destination addresses, port information, and protocol type) are extracted. This involves implementing a hardware decoder on the FPGA to parse various protocol layer data in Ethernet frames, such as the Ethernet header, IP header, and TCP / UDP header. By implementing data stream parsing in hardware, the FPGA can parse each field of a data packet within a clock cycle, minimizing latency.

[0055] For example, based on the information in the packet header, the FPGA classifies the traffic through a hash table or a lookup table, and packets of different categories can be stored in different memory areas (such as DDR or FIFO).

[0056] S120, based on task requirements and network characteristics, extracting traffic features from the data packet sequence to build a data fitting model.

[0057] In some embodiments, extracting traffic features from the data packet sequence based on task requirements and network characteristics to construct a data fitting model includes:

[0058] S121, extracting traffic features from the data packet sequence based on task requirements and network characteristics, and performing fusion calculation on each traffic feature to generate a traffic feature vector.

[0059] In some embodiments, the traffic characteristics include length distribution, arrival time interval, burst rate and periodicity characteristics of data packets.

[0060] For example, the above-mentioned traffic feature extraction operation can be implemented in FPGA. The specific extraction steps are as follows:

[0061] (1) Data packet length extraction includes: FPGA directly extracts the packet length information from the Ethernet frame header through bit operations. The length of the data packet can usually be found in the "Length / Type" field of the Ethernet frame, or calculated from the starting offset between the packet header and the data part.

[0062] Use hardware counters to count the frequency of packets of different lengths. For example, the FPGA can divide the packet length values into different intervals (such as less than 128 bytes, 128-512 bytes, etc.) and use memory or registers to store the number of packets in each interval.

[0063] Use the FPGA's BRAM (block RAM) or FIFO queue to store packet length information and read it in parallel to ensure that the length information of each packet can be processed in real time. Use parallel adders and comparators (such as an adder tree structure) to accelerate statistical operations in the FPGA.

[0064] (2) Calculating the time interval between adjacent data packets includes: FPGA uses its built-in high-precision clock (such as a nanosecond clock) to assign a timestamp to each data packet, and each time the FPGA captures a data packet, it records the current timestamp.

[0065] Whenever a new data packet is captured, the FPGA subtracts the current timestamp from the timestamp of the previous packet to obtain the time interval (i.e., the packet arrival interval). A high-precision timer (such as the built-in timer in the FPGA or a high-precision external clock) is used to ensure nanosecond-level accuracy.

[0066] Time intervals can be directly stored in the FPGA's BRAM for subsequent analysis. To reduce data throughput, time intervals can be calculated by batch or by stream. Time intervals can be grouped into different intervals (such as less than 100 nanoseconds, 100-500 nanoseconds, etc.) and statistically analyzed to analyze the timing characteristics of network traffic.

[0067] (3) Extracting the burst rate involves detecting the arrival time of packets and calculating the number of packets arriving per second or the number of bytes transmitted per second. For example, if 100 packets arrive in 1 second, the packet arrival rate is 100 packets / second.

[0068] A timer (e.g., a 1-second window) is used to calculate the traffic rate per unit time and detect traffic bursts. For example, if the packet arrival rate is much higher than the normal traffic rate within a short period of time, the FPGA can mark it as a "burst."

[0069] Traffic is classified by burst size (e.g., rate above a certain threshold), and the duration and burst intensity (i.e., rate) of each burst are recorded. Utilizing FPGA hardware acceleration, when traffic reaches the burst threshold, traffic shaping or priority adjustment strategies are triggered to avoid network congestion.

[0070] For example, various traffic features (such as packet length, arrival interval, burst rate, etc.) are integrated into a traffic feature vector to facilitate subsequent analysis, specifically including:

[0071] The characteristics of each data packet (such as packet length, arrival interval, burst rate, etc.) can be combined into a traffic feature vector. Each traffic feature vector contains multiple dimensions, representing different data packet characteristics (for example: [packet length, arrival interval, burst rate]).

[0072] FPGA uses parallel computing units to parallelize and aggregate the traffic feature vectors of multiple data packets in real time. Through efficient registers and hardware accelerators, these features are quickly aggregated (such as average, variance, maximum value, etc.) and the traffic feature vectors are output to the downstream intelligent analysis module for traffic pattern recognition, anomaly detection and prediction.

[0073] S122: Based on the flow characteristics, a statistical regression method is used to fit the flow change trend, and a Fourier transform algorithm is used to detect the periodic characteristics of the flow.

[0074] For example, a simple Fourier transform (FFT) or sliding window algorithm is used to detect the periodic characteristics of the traffic flow, wherein the periodic traffic flow is generally manifested as a regular change in the fluctuation amplitude of the traffic flow within a fixed time interval.

[0075] Based on the sliding window mechanism, the traffic flow changes are calculated within each fixed time window, and fast Fourier transform (FFT) or autocorrelation calculation is performed through hardware acceleration to find periodic features. The identified periodic information (cycle length, amplitude, etc.) is stored in the FPGA memory to provide a reference for subsequent intelligent analysis modules.

[0076] S123 , based on the traffic change trend and the periodic characteristics of the traffic, a probability distribution algorithm is used to model the traffic feature vector to generate a data fitting model.

[0077] For example, the choice of data fitting model should be based on the characteristics of the data and the analysis requirements. Common models include linear regression, nonlinear regression, time series analysis models (such as ARIMA), machine learning models (such as support vector machines, random forests) or deep learning models (such as LSTM, GRU), etc.

[0078] For example, the traffic data can be fitted based on a nonlinear regression model to estimate the changing trend of network traffic over time, specifically including:

[0079] First, the Newton method or quasi-Newton method (such as BFGS) is used to solve the parameters in the nonlinear model. Second, when implementing the nonlinear optimization process in an FPGA, data needs to be input in batches, and the sum of squared errors (SSE) of the model needs to be calculated in parallel within each batch. Finally, the FPGA's parallel computing unit is used to calculate the gradient of each iteration and update the model parameters. Parallel computing includes parallelizing the SSE calculation, gradient calculation, and parameter update steps to significantly improve computational efficiency. Hardware accelerators (such as dedicated multipliers and adders) are used to calculate gradients and reduce delays in the iterative process. The resulting nonlinear regression model (i.e., data fitting model) is used for future traffic prediction.

[0080] For example, a linear regression model (y=ax+b) may be used to fit traffic data to estimate the trend of network traffic over time, specifically including:

[0081] First, regression modeling is performed based on traffic feature vectors. Hardware is used to collect data packet information to generate time series or traffic series. Second, the FPGA's parallel computing units are used to calculate the parameters of the regression model, and the linear regression model is optimized using gradient descent. At each clock cycle, the FPGA calculates and updates the regression coefficients (a, b). Finally, matrix operations are used to solve the regression problem. The FPGA also uses parallel adders and multipliers to solve the matrix to improve computational efficiency. For multiple data points, matrix multiplication can be used to accelerate the calculation formula:

[0082] (X T *X) -1 *X T *Y

[0083] Where X is the input feature matrix and Y is the output target value.

[0084] For example, the built-in adders, multipliers and storage units (such as BRAM and FIFO) of the FPGA are used to perform matrix operations in parallel, which utilizes a pipeline structure to parallelize different data calculation steps, thereby reducing delays.

[0085] The calculated regression parameters can be stored in FPGA internal registers or external memory (such as DDR), and the subsequent traffic prediction and optimization modules can read these parameters.

[0086] S130 , fitting the data packet sequence based on the data fitting model to generate fitting time series data.

[0087] In some embodiments, fitting the data packet sequence based on the data fitting model to generate fitting time series data includes:

[0088] S131, fitting the mean data and the fluctuation data in the traffic feature vector based on the data fitting model to generate mean time series data and fluctuation time series data;

[0089] For example, based on the established data fitting model, the mean data and fluctuation data in the traffic feature vector are fitted respectively, wherein the mean data reflects the overall average level of network traffic, while the fluctuation data reveals the degree of change and stability of traffic.

[0090] For example, for mean data, we can generate mean time series data by fitting a model to predict its trend over time. This process helps capture long-term behavioral patterns of network traffic, such as periodic fluctuations, growth trends, or decay trends.

[0091] For fluctuating data, we also apply fitting models to analyze its changing patterns and generate fluctuating time series data. Fluctuating time series data can reveal short-term volatility and abnormal events in network traffic, such as sudden traffic and network attacks.

[0092] S132, synthesize the mean time series data and the fluctuation time series data according to the time series correspondence to generate fitting time series data.

[0093] For example, the mean time series data and the fluctuation time series data are synthesized according to their corresponding time series to generate the final fitted time series data. This step ensures the integrity and consistency of the data, facilitating subsequent applications such as network traffic analysis, prediction, and anomaly detection. Furthermore, comprehensive analysis of the fitted time series data can provide a deeper understanding of the dynamic behavior of network traffic and improve the intelligent level of network management.

[0094] S140, performing cluster analysis on the fitted time series data, predicting future network traffic change trends, and acquiring corresponding preset strategies based on the prediction results.

[0095] In some embodiments, performing cluster analysis on the fitted time series data, predicting future network traffic change trends, and obtaining corresponding preset strategies based on the prediction results include:

[0096] S141, performing cluster analysis on the fitted time series data based on a clustering algorithm, dividing the data into different clusters, wherein each cluster represents a traffic pattern and / or traffic change trend;

[0097] For example, traffic data is divided into multiple clusters based on a clustering algorithm, each cluster represents a specific traffic pattern or trend, and time series data is extracted from each cluster. These data may be sampled in hours, days, weeks or months.

[0098] For example, the initial cluster centers are provided by the CPU or randomly selected from the data, stored in the FPGA internal register, and the parallel computing unit is used to calculate the distance from each data point to all cluster centers:

[0099] Euclidean distance formula:

[0100]

[0101] The multiplier and adder are used in parallel to implement the square sum operation in the FPGA, and the distance calculation result is stored in the BRAM or FIFO of the FPGA.

[0102] For example, for each data point, the nearest cluster center is selected. The FPGA uses a comparator to directly compare the distance between each data point and the cluster center, assigns the point to the nearest cluster, and calculates the mean of the new points in each cluster as the new cluster center:

[0103] New center formula:

[0104]

[0105] The FPGA's internal pipeline structure calculates the cumulative sum of each cluster in parallel. After the cluster center is updated, it is stored in a register or memory. The FPGA controller then checks whether the cluster center has converged (the change before and after the update is less than a threshold). If not, the distance calculation and cluster assignment steps are repeated.

[0106] Exemplarily, the FPGA uses parallel units to calculate the similarity (such as Euclidean distance or cosine similarity) of each pair of data points and stores the similarity matrix in the BRAM of the FPGA.

[0107] Starting from the two most similar points or clusters, the FPGA uses a priority queue to store similarity values and quickly find the pair with the greatest similarity.

[0108] After merging, the similarity matrix is updated (the similarity between the new cluster and other points is recalculated), and the FPGA uses parallel units to update multiple similarity values at the same time.

[0109] S142, performing trend analysis on the traffic data in each cluster based on the time series prediction model, and combining the analysis results of each cluster to predict future traffic change trends;

[0110] Exemplarily, select a suitable time series prediction model: according to the characteristics of the data and the prediction requirements, select a suitable time series prediction model, among which commonly used models include ARIMA (autoregressive integrated moving average) model, exponential smoothing method, LSTM (long short-term memory) neural network, etc.

[0111] A continuous time period is selected from the historical traffic data as the training set to ensure that the time period contains various typical trends and patterns of traffic data, such as daily fluctuations, seasonal changes, and emergencies. Based on the characteristics of the traffic data, appropriate labels (such as traffic type, traffic level, etc.) are defined and each training sample is labeled so that these labels can be used for supervised learning during the training process. The target variable is determined according to the task requirements, such as predicting future traffic values and traffic change trends.

[0112] The selected model is trained based on the training set to determine the model parameters; the prediction performance of the model is evaluated through cross-validation or other validation methods to ensure that the model can accurately capture the trends and patterns of traffic data.

[0113] First, the traffic data in each cluster is predicted based on the trained model to obtain the traffic trend in the future period; second, based on the prediction results, the long-term trend of the traffic data (such as increase, decrease or stability), seasonal changes (such as periodic fluctuations) and other potential patterns (such as sudden traffic, abnormal fluctuations, etc.) are identified.

[0114] Finally, based on the results of trend analysis, the characteristics and behavior patterns of traffic data in each cluster are interpreted, and the analysis results are applied to the prediction and optimization decisions of industrial TSN network traffic, such as adjusting network bandwidth, implementing traffic shaping strategies, or optimizing network configuration based on the prediction results.

[0115] For example, for ARIMA and LSTM models, FPGA hardware resources (such as adders, multipliers, and memory) are used to accelerate steps such as autoregression, moving average, and matrix operations. The data flow control, pipeline structure, and parallel processing implemented in the FPGA ensure efficient and real-time time series modeling and forecasting.

[0116] For example, the prediction results (future flow values) obtained through time series modeling can be output to external systems for further flow control and optimization. The FPGA stores the prediction results in local memory and transmits them to downstream systems via DMA or high-speed interfaces (such as PCIe and Ethernet).

[0117] S143: Obtain a corresponding preset strategy from a preset strategy library based on the prediction result.

[0118] S150, dynamically adjusting network configuration based on the prediction result and the corresponding preset strategy, performing traffic shaping and priority scheduling.

[0119] For example, first, analyze the prediction results to identify possible traffic peaks, troughs and abnormal traffic events; and formulate a series of preset strategies based on business needs and network characteristics. These strategies should cover aspects such as traffic shaping, priority scheduling, and bandwidth allocation. Among them, the preset strategies may include traffic shaping rules based on time periods (such as different strategies for weekdays and weekends, daytime and nighttime), priority scheduling rules based on application types (such as high priority for real-time audio and video applications, low priority for file transfer applications), etc.

[0120] Secondly, based on the prediction results and preset strategies, the configuration parameters of network devices are dynamically adjusted, such as the queue management settings of routers and the flow control strategies of switches. Through traffic shaping technology, network traffic is limited to a predetermined rate range to avoid network congestion and delays. With priority scheduling technology, different processing priorities are assigned to different traffic flows based on their importance and urgency, ensuring the smooth operation of critical services.

[0121] Finally, after implementing dynamic adjustments, the network traffic in the industrial TSN network is monitored in real time to observe the adjustment effect. Based on the monitoring results, the preset strategies and network configurations are adjusted in a timely manner to cope with changes in network conditions and adjustments to business needs.

[0122] In some embodiments, the method further comprises:

[0123] Encrypt sensitive data and traffic information during data transmission and control user access rights to data.

[0124] For example, TLS / SSL protocols are used to encrypt transport layer data to ensure that data is not eavesdropped, tampered with, or leaked during transmission. All transmission of sensitive data, including but not limited to user information, transaction records, and business data, should be protected via HTTPS or Secure Sockets Layer.

[0125] For example, additional encryption measures can be implemented at the application layer, such as using a symmetric encryption algorithm (such as AES) to encrypt the storage and transmission of sensitive data, ensuring that even if the data is intercepted during transmission, it cannot be easily decrypted.

[0126] For example, metadata in network traffic (such as IP addresses, port numbers, transmission time, etc.) can also be obfuscated or encrypted to prevent attackers from inferring sensitive data or network structure by analyzing traffic information.

[0127] For example, you can also implement a strong password policy, requiring users to use complex and regularly changed passwords, and adopt multi-factor authentication (MFA), such as combining passwords, mobile phone verification codes, biometric recognition, etc., to improve account security.

[0128] In some embodiments, the method further comprises:

[0129] Conduct real-time detection of network traffic during data transmission, and take filtering and control measures for abnormal traffic.

[0130] For example, anomaly detection aims to quickly identify abnormal behaviors (such as attacks and congestion) from network traffic. This is an important part of network performance optimization and security management and can be achieved through the following methods:

[0131] (1) Statistical feature analysis method: extract statistical features, including the mean and variance of the message arrival interval, traffic rate (bytes or packets per second), and packet length distribution; use the FPGA parallel computing module to calculate these features in real time, set thresholds, and calculate the threshold range of each feature based on historical normal traffic data; compare the current feature value with the threshold in real time: if it exceeds the threshold range, it is marked as abnormal, triggering an alarm signal, notifying the upper system through interruption or DMA transmission, and storing the abnormal detection results in the FPGA internal memory (such as FIFO).

[0132] (2) Rule detection method: pre-define abnormal rules, including: traffic exceeding a specific number of bytes per second, the number of packets with the same source IP exceeding a threshold, etc., and store the predetermined rules in the form of a lookup table in the FPGA; FPGA extracts key information (such as source IP, destination IP, protocol type) by parsing the data packet, and compares in real time whether the traffic rate per second exceeds the set value; if it exceeds the set value, the abnormal traffic is directly marked, or the control module is triggered (such as discarding abnormal packets, limiting the speed).

[0133] (3) Frequency domain analysis method: perform periodic sampling of traffic data and store the sampled values in the FIFO or BRAM of the FPGA; perform frequency domain conversion on the sampled data based on the built-in FFT accelerator of the FPGA and analyze abnormal features in the spectrum (such as abnormal high frequency or periodic peaks); extract the amplitude and frequency features of the spectrum, compare them with the frequency distribution of normal traffic, set a frequency threshold, and mark the frequency features that exceed the threshold as abnormal. Abnormal traffic usually appears as a specific high-frequency component on the spectrum, and the results are stored or directly transmitted to the alarm module.

[0134] In some embodiments, the method further comprises:

[0135] Visualize the overall status of network traffic and traffic prediction results during data transmission.

[0136] For example, the user-friendly visual graphical interface supports a variety of visual chart types (such as line charts, bar charts, pie charts, heat maps, etc.) to intuitively display the overall status of network traffic; it can display key indicators of current network traffic in real time, such as total traffic, bandwidth utilization, traffic distribution of different protocols, traffic of major source / destination addresses, etc.

[0137] It also provides a historical data comparison function, which can display the historical change trend of network traffic through a timeline or sliding window.

[0138] Traffic forecast results can also be displayed in the form of prediction curves or prediction intervals to intuitively present the expected changes in future network traffic.

[0139] Finally, it also supports custom alarm rules. When the network traffic reaches the preset threshold, an alarm message will be sent to the network administrator through the graphical interface or text message, email, etc.

[0140] According to the embodiments of the present disclosure, the present disclosure first realizes the intelligent prediction of industrial TSN network traffic by integrating machine learning algorithms, providing a basis for subsequent traffic control and optimization; secondly, based on the prediction results and preset strategies, it implements refined traffic shaping and priority scheduling to ensure the efficient utilization of industrial TSN network resources; thirdly, by providing an intuitive visual traffic monitoring interface, it facilitates administrators to quickly respond to network problems, thereby improving the efficiency and convenience of network management; finally, by encrypting sensitive data and traffic information and implementing strict access control policies, it ensures the security of devices and the privacy of user data.

[0141] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present disclosure is not limited by the order of the actions described, because according to the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present disclosure.

[0142] The above is an introduction to the method embodiment. The following is a further explanation of the solution disclosed in the present disclosure through an apparatus embodiment.

[0143] Figure 2 FIG shows a structural diagram of a traffic prediction and dynamic scheduling device for an industrial TSN network provided by an embodiment of the present disclosure, as shown in FIG. Figure 2 As shown, a traffic prediction and dynamic scheduling device 200 for an industrial TSN network may include:

[0144] The data acquisition module 210 is used to collect data from the industrial TSN network and perform preprocessing to generate a data packet sequence based on time series sorting;

[0145] A model building module 220 is configured to extract traffic features from the data packet sequence and build a data fitting model based on task requirements and network characteristics;

[0146] A data fitting module 230 is configured to fit the data packet sequence based on the data fitting model to generate fitting time series data;

[0147] The analysis and prediction module 240 is used to perform cluster analysis on the fitted time series data, predict future network traffic change trends, and obtain corresponding preset strategies based on the prediction results;

[0148] The adjustment and optimization module 250 is used to dynamically adjust the network configuration based on the prediction results and the corresponding preset strategies, and perform traffic shaping and priority scheduling.

[0149] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0150] In the technical solutions disclosed herein, the acquisition, storage, and application of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0151] Figure 3 A block diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. Electronic device 300 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 300 may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0152] like Figure 3 As shown, the electronic device 300 may include a computing unit 301, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 302 or a computer program loaded from a storage unit 308 into a random access memory (RAM) 303. Various programs and data required for the operation of the electronic device 300 may also be stored in the RAM 303. The computing unit 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0153] Multiple components in the electronic device 300 are connected to the I / O interface 305, including an input unit 306, such as a keyboard, a mouse, etc.; an output unit 307, such as various types of displays, speakers, etc.; a storage unit 308, such as a magnetic disk, an optical disk, etc.; and a communication unit 309, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 309 allows the electronic device 300 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0154] The computing unit 301 can be a variety of general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 301 performs the various methods and processes described above, such as method 100. For example, in some embodiments, method 100 can be implemented as a computer program product, including a computer program tangibly embodied in a computer-readable medium, such as a storage unit 308. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 300 via ROM 302 and / or communication unit 309. When the computer program is loaded into RAM 303 and executed by the computing unit 301, one or more steps of the method 100 described above can be performed. Alternatively, in other embodiments, the computing unit 301 can be configured to perform method 100 in any other appropriate manner (e.g., by means of firmware).

[0155] The various embodiments described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chip (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system comprising at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0156] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0157] In the context of the present disclosure, a computer-readable medium can be a tangible medium that can contain or store a program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a computer-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0158] It should be noted that the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute method 100 and achieve the corresponding technical effects achieved by executing the method in the embodiment of the present disclosure. For the sake of brevity, they will not be repeated here.

[0159] In addition, the present disclosure also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the method 100 is implemented.

[0160] To provide interaction with a user, the embodiments described above may be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0161] The embodiments described above can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with embodiments of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0162] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0163] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0164] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A traffic prediction and dynamic scheduling method for industrial TSN networks, characterized in that: include: Collect and pre-process data from industrial TSN networks to generate data packet sequences based on time series sorting; Based on task requirements and network characteristics, traffic features are extracted from the data packet sequence to build a data fitting model; Fitting the data packet sequence based on the data fitting model to generate fitting time series data; Performing cluster analysis on the fitted time series data to predict future network traffic change trends, and obtaining corresponding preset strategies based on the prediction results; Based on the prediction results and corresponding preset strategies, the network configuration is dynamically adjusted to perform traffic shaping and priority scheduling.

2. The method according to claim 1, wherein The extracting traffic features from the data packet sequence based on task requirements and network characteristics to construct a data fitting model includes: Extracting traffic features from the data packet sequence based on task requirements and network characteristics, and performing fusion calculation on each traffic feature to generate a traffic feature vector; Based on the traffic characteristics, a statistical regression method is used to fit the traffic change trend, and a Fourier transform algorithm is used to detect the periodic characteristics of the traffic; Based on the flow variation trend and the periodic characteristics of the flow, a probability distribution algorithm is used to model the flow feature vector to generate a data fitting model.

3. The method according to claim 2, wherein The step of fitting the data packet sequence based on the data fitting model to generate fitting time series data includes: Fitting the mean data and the fluctuation data in the traffic feature vector based on the data fitting model to generate mean time series data and fluctuation time series data; The mean time series data and the fluctuation time series data are synthesized according to the time series correspondence to generate fitting time series data.

4. The method according to claim 2, wherein The traffic characteristics include the length distribution, arrival time interval, burst rate and periodicity characteristics of data packets.

5. The method according to claim 1, wherein The cluster analysis of the fitted time series data to predict future network traffic change trends and obtaining corresponding preset strategies based on the prediction results includes: Performing cluster analysis on the fitted time series data based on a clustering algorithm to divide the data into different clusters, wherein each cluster represents a traffic pattern and / or traffic change trend; Based on the time series prediction model, the traffic data in each cluster is analyzed separately, and the analysis results of each cluster are combined to predict the future traffic change trend; Based on the prediction results, the corresponding preset strategy is obtained from the preset strategy library.

6. The method according to claim 1, wherein The method further comprises: Encrypt sensitive data and traffic information during data transmission and control user access rights to data.

7. The method according to claim 1, wherein The method further comprises: Conduct real-time detection of network traffic during data transmission, and take filtering and control measures for abnormal traffic.

8. The method according to claim 1, wherein The method further comprises: Visualize the overall status of network traffic and traffic prediction results during data transmission.

9. A traffic prediction and dynamic scheduling device for industrial TSN networks, characterized in that: include: The data acquisition module is used to collect and pre-process data from the industrial TSN network to generate a sequence of data packets based on time series sorting; A model building module is used to extract traffic features from the data packet sequence and build a data fitting model based on task requirements and network characteristics; A data fitting module, configured to fit the data packet sequence based on the data fitting model to generate fitting time series data; An analysis and prediction module is used to perform cluster analysis on the fitted time series data, predict future network traffic change trends, and obtain corresponding preset strategies based on the prediction results; The adjustment and optimization module is used to dynamically adjust the network configuration based on the prediction results and the corresponding preset strategies, and perform traffic shaping and priority scheduling.

10. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 8.

Images