Wireless network environment scenario classification and sub-classification method, system and storage medium

Abstract

The application discloses a wireless network environment scene classification and sub-classification method and system and a storage medium. The number of BK frames, BE frames, VI frames and VO frames in a medium access control layer and the average value of a physical layer air interface utilization rate of a target wireless network environment in a set time interval are acquired, and a corresponding classifier model is introduced for scene classification to obtain a scene classification result of the target wireless network environment. A high-dimensional time sequence data set is constructed according to the number of BK frames, the number of BE frames, the number of VI frames, the number of VO frames, the number of Data frames, the number of BA frames, the number of RTS frames, the number of CTS frames, the total number of frames, the average value of the physical layer air interface utilization rate and the total length and retransmission rate of a Payload in the set time interval, and a TICC algorithm is used for calculation to obtain a sub-scene classification result of the target wireless network environment. The application can effectively solve the problems of scene classification and sub-scene classification of a wireless network environment and greatly improve the performance and efficiency of wireless network environment optimization.

Description

Technical Field

[0001] This invention belongs to the field of wireless network technology, specifically relating to wireless network environment scenario classification and sub-classification methods, systems, and storage media. Background Technology

[0002] With the rapid development of Wireless Local Area Network (WLAN) technology, Wi-Fi devices have gradually become the preferred option for accessing the wireless internet due to their excellent communication performance, ease of deployment, and cost-effectiveness. Simultaneously, the increasing demand for high-quality services such as VR / AR and metaverse is placing higher demands on the performance parameters of Wi-Fi devices. Although each generation of the IEEE 802.11 protocol has optimized the physical (PHY) and media access control (MAC) layers, the best solution to date remains deploying more access points (APs). This inevitably leads to an excessive number of nodes in the scenario, making the wireless network environment increasingly complex. Optimizing the wireless network environment is therefore urgent, and precise scenario classification and sub-scenario classification can greatly improve the performance and efficiency of wireless network environment optimization.

[0003] In different scenarios, the deployment strategies and number of access points (APs) generally differ, and the mean and peak values ​​of user stations (STAs) over time also generally vary. This results in different wireless network characteristics across different scenarios. Attempting to describe the characteristics of wireless networks in all scenarios using a large and general model is difficult and ineffective. However, classifying scenarios beforehand greatly reduces the difficulty of analyzing and optimizing individual scenarios. Therefore, wireless network scenario classification is of great significance.

[0004] In actual testing, it was found that in scenarios such as cafeterias, shopping malls, and arcades, the changes in the wireless network environment are mainly affected by the number of STA nodes, exhibiting strong temporal fluctuations: the wireless network characteristics during peak and off-peak hours differ significantly within the same scenario, while the characteristics during peak or off-peak hours in different scenarios show some similarities. Besides periodic temporal fluctuations, unpredictable sudden events can also cause drastic changes in the wireless network environment characteristics, making the wireless network characteristics of the same scenario vastly different before and after the event. Ignoring the differences in the temporal dimension and using a single label to describe a wireless network environment is inappropriate; sub-scenario classification is necessary based on scenario classification.

[0005] Existing methods for analyzing and classifying wireless network environments are mostly traditional models driven by expert knowledge, such as Markov models. These models are validated by setting very strong assumptions and combining them with simulation datasets. However, with the rapid development of WLAN technology and the increasing complexity of wireless network environments, these unchanging and costly traditional models lack scientific rigor and adaptability when faced with the enormous challenges posed by newly emerging complex wireless network environments. They also fail to extract and generalize new knowledge from new data. Summary of the Invention

[0006] The purpose of this invention is to provide a method, system, and storage medium for classifying and subclassifying wireless network environment scenarios, in order to solve the aforementioned problems existing in the prior art.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] Firstly, it provides methods for classifying and subclassifying wireless network environment scenarios, including:

[0009] Obtain media access control layer frame information and physical layer air interface utilization information of the target wireless network environment;

[0010] Calculate the number of BK frames, BE frames, VI frames, and VO frames in the Media Access Control (MAC) layer frame information within a set time interval, and calculate the average physical layer air interface utilization rate within the set time interval based on the physical layer air interface utilization rate information.

[0011] The number of BK frames, BE frames, VI frames, VO frames, and the average physical layer air interface utilization within a set time interval are imported into a preset classifier model for scene classification to obtain the scene classification results of the target wireless network environment.

[0012] Calculate the number of Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the total length of the Payload, and the retransmission rate in the Media Access Control (MAC) layer frame information within a set time interval.

[0013] A high-dimensional time series dataset is constructed based on the number of BK frames, BE frames, VI frames, VO frames, Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the average physical layer air interface utilization, and the total length and retransmission rate of the payload within a set time interval.

[0014] The high-dimensional time series dataset is imported into the pre-built Time Series Separation and Clustering (TICC) model to calculate the high-dimensional time series segmentation and clustering results. These results are then used as the sub-scene classification results for the target wireless network environment.

[0015] In one possible design, the pre-built classifier model is a tested KNN model.

[0016] In one possible design, the testing process for the KNN model includes:

[0017] Construct the initial KNN model;

[0018] Obtain sample data sets for several wireless network environment scenarios to form the original dataset;

[0019] Use 70% of the original dataset as the training set and the remaining 30% as the test set;

[0020] The initial KNN model was tested using the training and test sets to obtain the tested KNN model.

[0021] In one possible design, the pre-built classifier model is a trained CART model.

[0022] In one possible design, the training process of the CART model includes:

[0023] Build the initial CART model;

[0024] Obtain sample data sets for several wireless network environment scenarios to form the original dataset;

[0025] Use 70% of the original dataset as the training set and the remaining 30% as the test set;

[0026] The initial CART model was tested and trained using the training and test sets to obtain the trained CART model.

[0027] In one possible design, the pre-built classifier model is a trained SVM model.

[0028] In one possible design, the training process of the SVM model includes:

[0029] Construct the initial SVM model;

[0030] Obtain sample data sets for several wireless network environment scenarios to form the original dataset;

[0031] Use 70% of the original dataset as the training set and the remaining 30% as the test set;

[0032] The initial SVM model is tested and trained using the training and test sets to obtain the trained SVM model.

[0033] In one possible design, importing the high-dimensional time series dataset into a pre-defined Time Series Separation and Clustering (TICC) model to calculate the high-dimensional time series segmentation and clustering results includes: setting the number of clusters and the sliding window length of the TICC model; importing the high-dimensional time series dataset into the TICC model with the set number of clusters and sliding window length for calculation to obtain the high-dimensional time series segmentation and clustering results.

[0034] Secondly, a wireless network environment scenario classification and sub-classification system is provided, including an acquisition unit, a first computing unit, a first classification unit, a second computing unit, a construction unit, and a second classification unit, wherein:

[0035] The acquisition unit is used to acquire media access control layer frame information and physical layer air interface utilization information of the target wireless network environment.

[0036] The first calculation unit is used to calculate the number of BK frames, BE frames, VI frames and VO frames in the Media Access Control layer frame information within a set time interval, and to calculate the average physical layer air interface utilization rate within the set time interval based on the physical layer air interface utilization rate information.

[0037] The first classification unit is used to input the number of BK frames, BE frames, VI frames, VO frames and the average physical layer air interface utilization within a set time interval into a preset classifier model for scene classification, so as to obtain the scene classification result of the target wireless network environment.

[0038] The second calculation unit is used to calculate the number of Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the total length of the Payload, and the retransmission rate in the Media Access Control layer frame information within a set time interval.

[0039] The construction unit is used to construct a high-dimensional time series dataset based on the number of BK frames, BE frames, VI frames, VO frames, Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the average physical layer air interface utilization, and the total length and retransmission rate of the payload within a set time interval.

[0040] The second classification unit is used to import the high-dimensional time series dataset into the preset Time Series Separation and Clustering (TICC) model, calculate the high-dimensional time series segmentation and clustering results, and use the high-dimensional time series segmentation and clustering results as the sub-scene classification results of the target wireless network environment.

[0041] Thirdly, a computer-readable storage medium is provided, on which instructions are stored, which, when executed on a computer, cause the computer to perform any of the methods described in the first aspect. Also provided is a computer program product containing instructions that, when executed on a computer, cause the computer to perform any of the methods described in the first aspect.

[0042] Beneficial Effects: This invention obtains the number of BK frames, BE frames, VI frames, and VO frames in the Medium Access Control (MAC) layer frame information of the target wireless network environment within a set time interval, as well as the average physical layer air interface utilization rate. This data is then imported into a corresponding classifier model for scene classification, yielding the scene classification results for the target wireless network environment. Furthermore, based on the number of BK frames, BE frames, VI frames, VO frames, Data frames, BA frames, RTS frames, CTS frames, the total number of frames, the average physical layer air interface utilization rate, and the total length and retransmission rate of the payload within the set time interval, a high-dimensional time series dataset is constructed. The TICC algorithm is then used to calculate the sub-scene classification results of the target wireless network environment based on this dataset. This effectively solves the scene classification and sub-scene classification problems of wireless network environments, significantly improving the performance and efficiency of wireless network environment optimization. Simultaneously, this invention uses a data-driven approach, avoiding the use of complex expert systems. Compared to traditional expert knowledge-driven models, it has lower operating costs and better generalization and adaptability. Attached Figure Description

[0043] 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.

[0044] Figure 1 This is a schematic diagram of the steps of the method in an embodiment of the present invention;

[0045] Figure 2 This is a schematic diagram of the dataset structure accepted by the TICC algorithm in this embodiment of the invention;

[0046] Figure 3 This is a schematic diagram of the clustering results of the TICC algorithm in an embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of the system configuration in an embodiment of the present invention;

[0048] Figure 5 This is a schematic diagram of the configuration of a computer device in an embodiment of the present invention. Detailed Implementation

[0049] It should be noted that the descriptions of these embodiments are intended to aid in understanding the invention and do not constitute a limitation thereof. The specific structural and functional details disclosed herein are merely for describing exemplary embodiments of the invention. However, the invention may be embodied in many alternative forms and should not be construed as being limited to the embodiments described herein.

[0050] It should be understood that the terms "first," "second," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance. Although the terms "first," "second," etc., may be used herein to describe various features, these features should not be limited by these terms. For example, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit, without departing from the scope of exemplary embodiments of the invention.

[0051] Specific details are provided in the following description to provide a complete understanding of the exemplary embodiments. However, those skilled in the art will understand that the exemplary embodiments can be implemented without these specific details. For example, the system may be shown in block diagrams to avoid obscuring the example with unnecessary details. In other embodiments, well-known processes, structures, and techniques may be shown without non-essential details to avoid obscuring the embodiments.

[0052] Example 1:

[0053] This embodiment provides a method for classifying and subclassifying wireless network environment scenarios, such as... Figure 1 As shown, the method includes the following steps:

[0054] S1. Obtain the Media Access Control (MAC) layer frame information and the Physical Layer air interface utilization information of the target wireless network environment.

[0055] In practice, the target wireless network environment can be initially collected using appropriate data acquisition equipment to obtain Media Access Control (MAC) frame information and Physical Layer (PHY) air interface utilization information. The MAC frame information contains most of the information transmitted between MAC layers in the wireless network environment at each moment during the equipment acquisition process.

[0056] S2. Calculate the number of BK frames, BE frames, VI frames, and VO frames in the Media Access Control (MAC) layer frame information within a set time interval, and calculate the average physical layer air interface utilization rate within the set time interval based on the physical layer air interface utilization rate information.

[0057] In practical implementation, considering the principle of the physical layer EDCA mechanism, the information features of BK frames, BE frames, VI frames, and VO frames in the MAC layer frame information are used to analyze real-world network scenarios. Since MAC layer frame information records information about a specific frame at a specific moment and is not suitable for direct use, the number of BK frames, BE frames, VI frames, and VO frames in the medium access control layer frame information within a set time interval can be calculated. For example, using a one-second interval, the number of BK frames, BE frames, VI frames, and VO frames in each second can be calculated as a feature. Simultaneously, the average physical layer air interface utilization within a set time interval, such as within one second, is also used as a feature to achieve the analysis and classification of the target wireless network environment.

[0058] S3. Import the number of BK frames, BE frames, VI frames, VO frames, and the average physical layer air interface utilization within the set time interval into a preset classifier model for scene classification to obtain the scene classification results of the target wireless network environment.

[0059] In practice, after obtaining the number of BK frames, BE frames, VI frames, VO frames, and the average physical layer air interface utilization rate within a set time interval, these numbers are used as features and imported into a pre-set classifier model for scene classification to obtain the scene classification results of the target wireless network environment.

[0060] The pre-set classifier model can be a tested KNN (K-Nearest Neighbors) model, CART (Classification and Regression Tree) model, or SVM (Support Vector Machine) model. The testing process for the KNN model, CART model, or SVM model includes:

[0061] First, construct the initial KNN model, CART model, or SVM model.

[0062] Then, sample data sets were collected for several wireless network environment scenarios to form the original dataset. For example, 5000 data sets were collected from each of the three scenarios: home, office, and retail. Each data set includes five parameters: the number of BK frames, BE frames, VI frames, VO frames, and the average air interface utilization rate within 1 second. Each data set is also labeled with a real-world scenario label, forming an original dataset with a capacity of 15000. The dataset structure is shown in Table 1 below.

[0063] Table 1 Original Dataset Structure

[0064]

[0065] Then, 70% of the original dataset is used as the training set, and the remaining 30% as the test set. The initial KNN model is tested using both the training and test sets, or the CART or SVM model is tested and trained to obtain the tested KNN model, or the tested and trained CART or SVM model. The model obtains predicted labels through training, and the true labels are compared with the predicted labels to calculate accuracy, precision, recall, and F1 score. The performance comparison of the KNN, CART, and SVM models for scene classification in a wireless network environment is shown in Table 2 below.

[0066] Table 2 Comparison of Scene Classification Performance of Three Classification Models

[0067]

[0068] As shown in Table 2, in the actual classification process, all three classifier models achieved a scene recognition accuracy of over 90%. Therefore, it is feasible to use the above classification models for scene recognition in wireless network environments.

[0069] S4. Calculate the number of Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the total length of the Payload, and the retransmission rate in the Media Access Control (MAC) layer frame information within the set time interval.

[0070] In practice, since sub-scene classification is more difficult than scene classification, more time-series features are added to the scene classification dataset to create a new dataset. Specifically, this involves calculating the number of Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the total payload length, and the retransmission rate within a set time interval in the Medium Access Control (MAC) frame information. Using Wireshark, the payload length of each frame can be explicitly provided, allowing the calculation of the total payload length within one second. Wireshark also clearly identifies whether each frame is a retransmission frame; therefore, the retransmission rate is obtained by calculating the number of retransmitted frames within one second and dividing by the total number of frames within that second.

[0071] S5. Construct a high-dimensional time series dataset based on the number of BK frames, BE frames, VI frames, VO frames, Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the average physical layer air interface utilization, and the total length and retransmission rate of the payload within a set time interval.

[0072] In practical implementation, the wireless network scenario sub-classification method is essentially a problem of segmenting and clustering the collected high-dimensional time series, which belongs to the high-dimensional time series subsequence clustering problem. Therefore, a high-dimensional time series dataset can be constructed based on the number of BK frames, BE frames, VI frames, VO frames, Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the average physical layer air interface utilization, and the total length and retransmission rate of the payload within a set time interval.

[0073] S6. Import the high-dimensional time series dataset into the pre-built time series segmentation and clustering algorithm TICC (ToeplitzInverse Covariance-Based Clustering) model to calculate the high-dimensional time series segmentation and clustering results, and use the high-dimensional time series segmentation and clustering results as the sub-scene classification results of the target wireless network environment.

[0074] In practice, after constructing the high-dimensional time series dataset, it can be imported into the pre-built Time Series Separation and Clustering (TICC) algorithm model for calculation, obtaining the high-dimensional time series segmentation and clustering results, which serve as the sub-scene classification results for the target wireless network environment. Unlike classical methodologies that only consider the shape similarity between the same feature, the TICC algorithm places greater emphasis on the correlation between different features and the overall structure of the high-dimensional time series. The TICC algorithm creates and maintains a Markov random field for each cluster, which demonstrates the interdependencies between features.

[0075] The TICC algorithm relies on the fundamental assumption that high-dimensional time series classified into the same category all possess the same unique, time-invariant "correlation" structure: regardless of where the sliding window of length w is truncated within these high-dimensional time series, the "correlation" structure of the windowed time series remains unchanged. To guarantee this time invariance, the TICC algorithm introduces a symmetric Toeplitz matrix structure, as shown below:

[0076]

[0077] For a high-dimensional time series with n features, the TICC algorithm calculates the inverse covariance matrix θ of the high-dimensional time series of length w in the i-th cluster. Based on the above assumption, θ is a symmetric Toeplitz matrix, as shown below:

[0078]

[0079] Where A (0) A (1) ,…,A (w-1) ∈R n×n ,set up For A (k) The element in the m-th row and n-th column of the matrix, then This represents the dependency between the m-th time series feature at the current time and the n-th time series feature at the next k times. The TICC algorithm obtains the Markov random field for a cluster by using the inverse covariance matrix of the same cluster, and continuously updates the Markov random field while updating the inverse covariance matrix.

[0080] The ultimate goal of the TICC algorithm is to compute the inverse covariance matrix Θ = {Θ1, Θ2, ..., Θ3} of K clusters (the value of K needs to be set in advance). k The cluster affiliation of the timestamps is P = {P1, P2, ..., P} k},in This minimizes the following cost function:

[0081]

[0082] The TICC algorithm aims for sparse inverse covariance matrix Θ of each cluster to provide better interpretability; therefore, the L1 norm is introduced as a regularization term ||λΘ||. i ||1;

[0083] Meanwhile, the TICC algorithm aims to group adjacent time steps together as much as possible, hence the introduction of... Where β is a constant, when the previous time step and the current time step belong to the same cluster. The value is 0 otherwise it is 1;

[0084] The TICC algorithm uses maximum likelihood estimation to estimate parameters, therefore the middle term of the cost function is -ll(X). t ,Θ i Take the minimum value, ll(X) t ,Θ i The definition of ) is as follows:

[0085]

[0086] The optimization problem of the aforementioned cost function is a non-convex multi-objective problem. To achieve the optimal result, it is necessary to simultaneously optimize Θ = {Θ1, Θ2, ..., Θ}. k} and P = {P1, P2, ..., P k The two sets of parameters present a significant challenge. The TICC algorithm employs an alternating optimization strategy—first fixing one set of parameters, such as Θ = {Θ1, Θ2, ..., Θ...} k}, prioritize optimizing another set of parameters, for example, P = {P1, P2, ..., P} k When the cost function reaches a "temporary minimum", the parameter set P = {P1, P2, ..., P} is fixed. kInstead, optimize Θ = {Θ1, Θ2, ..., Θ} k The specific process optimization is as follows:

[0087] First, fix Θ={Θ1, Θ2,…,Θ k}, at this time ||λΘ i The value of ||1 is fixed and will not affect the optimization of the cost function. The original optimization problem is transformed into finding the optimal P = {P1, P2, ..., P}. k This makes the following expression take its minimum value:

[0088]

[0089] The optimization problem described above is similar to the second classic problem in Hidden Markov Fields, and can be solved using dynamic programming. Next, fix P = {P1, P2, ..., P...} k},at this time The value of is fixed and will not affect the optimization of the cost function. The original optimization problem is transformed into finding the optimal Θ = {Θ1, Θ2, ..., Θ}. k This makes the following expression take its minimum value:

[0090]

[0091] The optimization problem described above can be solved using the alternating direction method of multipliers (ADMM). When the aforementioned cost function reaches its minimum, the parameter set P = {P1, P2, ..., P...} is considered to be... k}、Θ={Θ1,Θ2,…,Θ k The optimal value has been reached, at which point P = {P1, P2, ..., P}. k This results in the segmentation and clustering of high-dimensional time series.

[0092] In practice, the TICC algorithm has been fully implemented in Python. It can accept high-dimensional time series datasets with N feature dimensions and T time series length, satisfying T >> N, as input. The aforementioned high-dimensional time series dataset is read into Python as a .csv file and transformed into... Figure 2 The data structure is shown below. Before using the TICC algorithm, the values ​​of the number of clusters K and the sliding window length w mentioned in the working principle need to be set in advance. At the start of the TICC algorithm, each point in the high-dimensional time series is randomly assigned to one of the K clusters, thus obtaining the parameter set P = {P1, P2, ..., P...}. k}; after obtaining the parameter set P = {P1, P2, ..., P} k After that, use the sliding window length w and such Figure 2The data shown is used to calculate the parameter set Θ = {Θ1, Θ2, ..., Θ} according to the method mentioned above in the working principle. k Using the alternating minimization method mentioned in the working principle, the optimal P = {P1, P2, ..., P} is obtained. k} and Θ={Θ1,Θ2,…,Θ k}, where P = {P1, P2, ..., P} k This represents the segmentation and clustering results of high-dimensional time series. The TICC algorithm handles real wireless network environment datasets well, and the clustering results at K=2 and K=3 have good interpretability, such as... Figure 3 The diagram shown is a schematic of the clustering results of the TICC algorithm when K=3.

[0093] Example 2:

[0094] This embodiment provides a wireless network environment scene classification and sub-classification system, such as Figure 4 As shown, it includes an acquisition unit, a first calculation unit, a first classification unit, a second calculation unit, a construction unit, and a second classification unit, wherein:

[0095] The acquisition unit is used to acquire media access control layer frame information and physical layer air interface utilization information of the target wireless network environment.

[0096] The first calculation unit is used to calculate the number of BK frames, BE frames, VI frames and VO frames in the Media Access Control layer frame information within a set time interval, and to calculate the average physical layer air interface utilization rate within the set time interval based on the physical layer air interface utilization rate information.

[0097] The first classification unit is used to input the number of BK frames, BE frames, VI frames, VO frames and the average physical layer air interface utilization within a set time interval into a preset classifier model for scene classification, so as to obtain the scene classification result of the target wireless network environment.

[0098] The second calculation unit is used to calculate the number of Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the total length of the Payload, and the retransmission rate in the Media Access Control layer frame information within a set time interval.

[0099] The construction unit is used to construct a high-dimensional time series dataset based on the number of BK frames, BE frames, VI frames, VO frames, Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the average physical layer air interface utilization, and the total length and retransmission rate of the payload within a set time interval.

[0100] The second classification unit is used to import the high-dimensional time series dataset into the preset Time Series Separation and Clustering (TICC) model, calculate the high-dimensional time series segmentation and clustering results, and use the high-dimensional time series segmentation and clustering results as the sub-scene classification results of the target wireless network environment.

[0101] Example 3:

[0102] This embodiment provides a computer device, such as... Figure 5 As shown, at the hardware level, it includes:

[0103] The data interface is used to establish data communication between the processor and the external wireless network access point in order to obtain relevant information about the target wireless network environment.

[0104] Memory, used to store instructions;

[0105] The processor is used to read instructions stored in the memory and execute the wireless network environment scenario classification and sub-classification method in Embodiment 1 according to the instructions.

[0106] Optionally, the computer system also includes an internal bus. The processor, memory, and display can be interconnected via the internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc.

[0107] The memory may include, but is not limited to, random access memory (RAM), read-only memory (ROM), flash memory, first-in-first-out (FIFO) memory, and / or first-in-last-out (FILO) memory. The processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0108] Example 4:

[0109] This embodiment provides a computer-readable storage medium storing instructions. When these instructions are executed on a computer, the computer performs the wireless network environment scenario classification and sub-classification method described in Embodiment 1. The computer-readable storage medium refers to a data storage medium, which may include, but is not limited to, floppy disks, optical disks, hard disks, flash memory, USB flash drives, and / or Memory Sticks. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable systems.

[0110] This embodiment also provides a computer program product containing instructions that, when executed on a computer, cause the computer to perform the wireless network environment scenario classification and sub-classification method in Embodiment 1. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable system.

[0111] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for classifying and subclassifying wireless network environment scenarios, characterized in that, include: Obtain media access control layer frame information and physical layer air interface utilization information of the target wireless network environment; Calculate the number of BK frames, BE frames, VI frames, and VO frames in the Media Access Control (MAC) layer frame information within a set time interval, and calculate the average physical layer air interface utilization rate within the set time interval based on the physical layer air interface utilization rate information. The number of BK frames, BE frames, VI frames, VO frames, and the average physical layer air interface utilization within a set time interval are imported into a preset classifier model for scene classification to obtain the scene classification results of the target wireless network environment. Calculate the number of Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the total length of the Payload, and the retransmission rate in the Media Access Control (MAC) layer frame information within a set time interval. A high-dimensional time series dataset is constructed based on the number of BK frames, BE frames, VI frames, VO frames, Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the average physical layer air interface utilization, and the total length and retransmission rate of the payload within a set time interval. The high-dimensional time series dataset is imported into the pre-built Time Series Segmentation and Clustering Algorithm (TICC) model to calculate the high-dimensional time series segmentation and clustering results. The high-dimensional time series segmentation and clustering results are then used as the sub-scene classification results for the target wireless network environment.

2. The wireless network environment scene classification and sub-classification method according to claim 1, characterized in that, The preset classifier model is the tested KNN model.

3. The wireless network environment scene classification and sub-classification method according to claim 2, characterized in that, The testing process for the KNN model includes: Construct the initial KNN model; Obtain sample data sets for several wireless network environment scenarios to form the original dataset; Use 70% of the original dataset as the training set and the remaining 30% as the test set; The initial KNN model was tested using the training and test sets to obtain the tested KNN model.

4. The wireless network environment scene classification and sub-classification method according to claim 1, characterized in that, The pre-set classifier model uses a trained CART model.

5. The wireless network environment scene classification and sub-classification method according to claim 4, characterized in that, The training process of the CART model includes: Build the initial CART model; Obtain sample data sets for several wireless network environment scenarios to form the original dataset; Use 70% of the original dataset as the training set and the remaining 30% as the test set; The initial CART model was tested and trained using the training and test sets to obtain the trained CART model.

6. The wireless network environment scene classification and sub-classification method according to claim 1, characterized in that, The pre-set classifier model uses a trained SVM model.

7. The wireless network environment scene classification and sub-classification method according to claim 6, characterized in that, The training process of the SVM model includes: Construct the initial SVM model; Obtain sample data sets for several wireless network environment scenarios to form the original dataset; Use 70% of the original dataset as the training set and the remaining 30% as the test set; The initial SVM model is tested and trained using the training and test sets to obtain the trained SVM model.

8. The wireless network environment scene classification and sub-classification method according to claim 1, characterized in that, The step of importing the high-dimensional time series dataset into the preset Time Series Segmentation and Clustering Algorithm (TICC) model to calculate the high-dimensional time series segmentation and clustering results includes: setting the number of clusters and the sliding window length of the TICC model; importing the high-dimensional time series dataset into the TICC model with the set number of clusters and sliding window length for calculation to obtain the high-dimensional time series segmentation and clustering results.

9. A wireless network environment scenario classification and sub-classification system, characterized in that, It includes an acquisition unit, a first calculation unit, a first classification unit, a second calculation unit, a construction unit, and a second classification unit, wherein: The acquisition unit is used to acquire media access control layer frame information and physical layer air interface utilization information of the target wireless network environment. The first calculation unit is used to calculate the number of BK frames, BE frames, VI frames and VO frames in the Media Access Control layer frame information within a set time interval, and to calculate the average physical layer air interface utilization rate within the set time interval based on the physical layer air interface utilization rate information. The first classification unit is used to input the number of BK frames, BE frames, VI frames, VO frames and the average physical layer air interface utilization within a set time interval into a preset classifier model for scene classification, so as to obtain the scene classification result of the target wireless network environment. The second calculation unit is used to calculate the number of Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the total length of the Payload, and the retransmission rate in the Media Access Control layer frame information within a set time interval. The construction unit is used to construct a high-dimensional time series dataset based on the number of BK frames, BE frames, VI frames, VO frames, Data frames, BA frames, RTS frames, CTS frames, the total number of all frames, the average physical layer air interface utilization, and the total length and retransmission rate of the payload within a set time interval. The second classification unit is used to import the high-dimensional time series dataset into the preset time series segmentation and clustering algorithm TICC model, calculate the high-dimensional time series segmentation and clustering results, and use the high-dimensional time series segmentation and clustering results as the sub-scene classification results of the target wireless network environment.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the method according to any one of claims 1-8.

Images