A method and device for switching frequency band of wireless terminal under FTTR equipment

Abstract

This application discloses a method and device for switching the frequency band of a wireless terminal under an FTTR device, the method comprising: continuously monitoring the signal strength of all wireless terminals connected to the 5G frequency band through the FTTR device; when the 5G signal strength of a target wireless terminal is detected to be lower than a preset threshold, the FTTR device simultaneously sends a BSS Transition Management Request message to the target wireless terminal through the 5G and 2.4G frequency bands; when the target wireless terminal receives at least one BSS Transition Management Request message, the network quality of the 5G and 2.4G frequency bands is evaluated through the FTTR device; the FTTR device determines whether to perform a switching operation on the target wireless terminal based on the network quality evaluation result. This application can make reasonable frequency band switching decisions for different scenarios.

Description

Technical Field

[0001] The present application belongs to the field of wireless terminal signal detection, and specifically relates to a method and device for switching frequency bands of wireless terminals under FTTR equipment. Background Art

[0002] In the prior art, when a user's wireless terminal is connected to the 5G frequency band of the FTTR device and moves to an area with weaker signals, it is usually necessary to trigger the frequency band switching to 2.4G through the BSS Transition Management Request message in the 802.11v protocol to maintain business continuity. However, this simple switching mechanism often ignores the complexity and dynamics of the network quality of different frequency bands, resulting in the possibility that the switching decision may not be accurate and timely. In addition, the prior art lacks an effective means to comprehensively evaluate the network quality, and cannot fully consider multiple factors such as signal strength, signal-to-noise ratio, bit error rate, real-time data throughput, transmission rate, packet loss rate, delay, and user behavior patterns, thereby limiting the intelligence of frequency band switching and the optimization of user experience. Summary of the invention

[0003] The main purpose of this application is to provide a method and device for switching frequency bands of wireless terminals under FTTR equipment, aiming to solve the problem of inaccurate and untimely frequency band switching in the prior art.

[0004] To achieve the above objectives, this application provides the following technical solutions:

[0005] A method for switching frequency bands of wireless terminals under FTTR equipment, comprising: continuously monitoring the signal strength of all wireless terminals connected under the 5G frequency band through the FTTR equipment; when it is monitored that the 5G signal strength of a target wireless terminal is lower than a preset threshold, the FTTR equipment simultaneously sends a BSS TransitionManagement Request message to the target wireless terminal through the 5G and 2.4G frequency bands; when the target wireless terminal receives at least one BSS TransitionManagement Request message, the network quality of the 5G and 2.4G frequency bands is evaluated; and based on the network quality evaluation result, it is determined whether to perform a switching operation on the target wireless terminal.

[0006] Optionally, the evaluation of the network quality of the 5G and 2.4G frequency bands includes: collecting network data of the 5G and 2.4G frequency bands, the network data including signal strength, signal-to-noise ratio, bit error rate, real-time data throughput, packet loss rate, average delay, application type and user behavior pattern; preprocessing the network data; constructing and training a network quality evaluation model; inputting the preprocessed network data into the trained network quality evaluation model to predict the probability of switching to the 2.4G frequency band.

[0007] Optionally, if the network quality assessment model predicts that the probability of switching to the 2.4G frequency band is higher than a preset threshold, the switching operation is performed; otherwise, the current 5G frequency band is maintained.

[0008] Optionally, the network data is preprocessed, including: normalizing the numerical data in the network data; one-hot encoding the categorical features in the network data; processing missing values ​​of the normalized and one-hot encoded network data; and enhancing the network data after the missing value processing.

[0009] Optionally, the network quality assessment model includes: an input layer, a feature engineering layer, a first hidden layer, a second hidden layer, a one-dimensional convolutional neural network, a recurrent neural network, an adaptive weighted fusion layer, a self-attention mechanism layer and an output layer connected in sequence.

[0010] Optionally, the first hidden layer includes: a first fully connected layer and a second fully connected layer, the second hidden layer includes a third fully connected layer and a fourth fully connected layer, and the first hidden layer and the second hidden layer both use residual connections.

[0011] Optionally, the network quality assessment model is trained through the following steps: collecting real network data to construct a network data set; dividing the network data set into a training set and a validation set in proportion; setting training parameters, training the model using the training set, and using the cross entropy loss function to calculate the loss of the current batch until the loss function converges; verifying the trained model using the validation set, and verifying the performance of the model using accuracy, recall rate, F1 score and mAP as evaluation indicators. When the evaluation indicators all reach 90%, the model is verified; otherwise, adjusting the training parameters or expanding the training samples to retrain the model until the model is verified.

[0012] In addition, the present application also provides a storage medium, which includes instructions, and when the instructions are executed on a computer, the computer executes the method as described in any of the preceding items.

[0013] In addition, the present application also provides an electronic device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in any of the preceding items when executing the program.

[0014] Compared with the prior art, this application can bring the following technical effects:

[0015] This application uses FTTR equipment to achieve intelligent switching of wireless terminals between 5G and 2.4G frequency bands. By continuously monitoring signal strength and sending switching requests when necessary, combined with the network quality assessment model to predict and decide whether to switch frequency bands, it optimizes network performance, reduces switching delays and packet loss, and improves user experience. This application improves the accuracy and robustness of network quality assessment by introducing advanced technologies such as feature engineering, adaptive weighted fusion, and self-attention mechanism, ensuring that reasonable frequency band switching decisions are made in different scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flowchart of a method for switching frequency bands of a wireless terminal under an FTTR device provided by an embodiment of the present application;

[0017] Figure 2 is a structural diagram of a network quality assessment model provided by another embodiment of the present application;

[0018] Figure 3 is a schematic diagram of the structure of a storage medium provided by an embodiment of the present application;

[0019] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0022] In the present invention, unless otherwise clearly specified and limited, the terms "connection", "fixation", etc. should be understood in a broad sense. For example, "fixation" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0023] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the meaning of "and / or" appearing in the full text includes three parallel schemes. Taking "A and / or B" as an example, it includes scheme A, or scheme B, or a scheme that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0024] Figure 1 FIG. 1 is a flow chart of a method for switching a frequency band of a wireless terminal under an FTTR device according to an embodiment of the present application, such as Figure 1 As shown, the method includes:

[0025] S100: Continuously monitors the signal strength of all wireless terminals connected to the 5G frequency band through FTTR equipment;

[0026] S200: When the 5G signal strength of a target wireless terminal is detected to be lower than a preset threshold, the FTTR device sends a BSS Transition Management Request message to the target wireless terminal through both the 5G and 2.4G frequency bands;

[0027] S300: After the target wireless terminal receives at least one BSS Transition Management Request message, the target wireless terminal evaluates the network quality of the 5G and 2.4G frequency bands through the FTTR device;

[0028] S400: The FTTR device determines whether to perform a handover operation on the target wireless terminal based on the network quality evaluation result.

[0029] This application can realize intelligent switching of wireless terminals between 5G and 2.4G frequency bands through FTTR equipment. By continuously monitoring signal strength and sending switching requests when necessary, it can optimize network performance, reduce switching delays and packet loss, and improve user experience.

[0030] In another exemplary embodiment, in step S300, the evaluating the network quality of the two frequency bands of 5G and 2.4G includes:

[0031] S301: Collect network data of 5G and 2.4G frequency bands, where the network data includes signal strength, signal-to-noise ratio, bit error rate, real-time data throughput, packet loss rate, average delay, application type, and user behavior pattern;

[0032] S302: Preprocessing the network data;

[0033] S303: Build and train a network quality assessment model;

[0034] S304: Inputting the pre-processed network data into the trained network quality assessment model to predict the probability of switching to the 2.4G frequency band.

[0035] In another exemplary embodiment, in step S302, preprocessing the network data includes:

[0036] S3021: Normalizing the numerical data in the network data;

[0037] In this step, Min-Max or Z-Score can be used for normalization. Through normalization, the numerical features can be scaled to [0, 1] or [-1, 1] to eliminate the dimensional differences between different features, accelerate model convergence, and prevent some features from dominating other features due to excessively large numerical ranges.

[0038] S3022: Perform one-hot encoding on the classification features in the network data;

[0039] In this step, for each classification feature, a binary vector is created with the same number of categories. For example, assuming that there are three application types: video streaming, web browsing, and file downloading, they can be encoded as: video streaming: [1, 0,0], web browsing: [0, 1, 0], and file downloading: [0, 0, 1]. By converting the classification features into binary vector form, the network quality assessment model can better process the classification data.

[0040] S3023: performing missing value processing on the normalized and one-hot encoded network data;

[0041] In this step, the missing values ​​in the data can be processed by deletion method, interpolation method, etc. to avoid affecting the training of the network quality assessment model.

[0042] S3024: Enhance the network data after the missing value processing.

[0043] In this step, the data can be enhanced by adding random noise or time shifting to increase the diversity of the data, thereby improving the generalization ability of the network quality assessment model.

[0044] In another embodiment, in step S304, Figure 2 As shown, the network quality assessment model includes an input layer, a feature engineering layer, a first hidden layer, a second hidden layer, a one-dimensional convolutional neural network (1D-CNN), a recurrent neural network (RNN), an adaptive weighted fusion layer, a self-attention mechanism layer and an output layer, which are connected in sequence.

[0045] In this embodiment, the input layer is used to input raw data, including signal strength (RSSI), signal-to-noise ratio (SNR), bit error rate (BER), real-time data throughput, average delay, packet loss rate, application type and user behavior pattern.

[0046] The feature engineering layer is used to extract the most helpful features for frequency band switching decisions from the input raw data to obtain more comprehensive network quality evaluation indicators. Specifically, the feature engineering layer includes the following specific operations:

[0047] 1. Calculate a comprehensive quality index (CQI), which is calculated based on multiple factors. The specific calculation is as follows:

[0048] in, w 1. w 2. w 3. w 4. w 5 and w 6 represent the adjustable weight coefficients, (Received Signal Strength Indicator) indicates the strength of the received signal. (Signal-to-Noise Ratio) indicates the signal-to-noise ratio. (BitError Rate) indicates the bit error rate. represents the probability of no bit errors, represents the real-time data throughput, Indicates delay, Indicates the packet loss rate, Indicates the probability of no packet loss.

[0049] 2. Calculate the Application Sensitivity Index (ASI). The ASI is used to define an index to measure the network quality requirements of a specific application. The specific calculation is as follows:

[0050] in, Indicates video streaming, Indicates online games, Indicates file transfer. Indicates web browsing, Indicates a voice call. a 1 means The weight coefficient of a 2 means The weight coefficient of a 3 means The weight coefficient of a 4 means The weight coefficient of a 5 means The weight coefficient of .

[0051] 3. Characterize the user behavior pattern (UBP). The UBP extracts the user's mobility pattern and usage habits by analyzing the user's historical behavior. The specific characterization is as follows:

[0052]

[0053] in, Indicates the mobile mode. Indicates the frequency of use, Indicates the usage time. Indicates the device type. b 1 means The weight coefficient of b 2 means The weight coefficient of b 3 means The weight coefficient of b 4 means The weight coefficient of .

[0054] It should be noted that this application introduces a feature engineering layer, which can achieve the following technical effects in the process of evaluating network quality:

[0055] 1. By calculating the comprehensive quality index (CQI), application sensitivity index (ASI) and user behavior pattern (UBP), the feature engineering layer can extract the most helpful features for frequency band switching decisions from the original data. These features not only include information on basic indicators such as signal strength (RSSI), signal-to-noise ratio (SNR), bit error rate (BER), but also combine multi-dimensional data such as real-time data throughput, average delay, packet loss rate, etc., to form a more comprehensive and relevant feature representation. This multi-dimensional feature combination can more accurately reflect the overall status of network quality, thereby improving the evaluation accuracy of the model.

[0056] 2. The CQI, ASI and UBP generated by the feature engineering layer have clear physical meanings, among which CQI comprehensively reflects the network quality, ASI represents the demand for network quality of specific applications, and UBP describes the user's usage habits. These features with clear meanings make the output of the network quality assessment model more interpretable. If the network quality assessment model performs poorly in certain scenarios, the problem can be located by analyzing these features, and the network quality assessment model can be optimized in a targeted manner.

[0057] 3. Through the processing of the feature engineering layer, the network quality assessment model no longer relies on a single indicator, but makes decisions based on multiple comprehensive indicators. This helps the network quality assessment model to maintain good performance under different environments and conditions. For example, even if a single indicator fluctuates abnormally, the network quality assessment model can still rely on other relevant features to make accurate judgments. In addition, the introduction of user behavior patterns (UBP) enables the network quality assessment model to better adapt to the dynamic changes of users, provide more personalized frequency band switching strategies, and further enhance the robustness and generalization ability of the network quality assessment model.

[0058] 4. The feature engineering layer can pre-process and combine the original features, reducing the number of features that need to be processed by the subsequent hidden layer, thereby simplifying the structure of the network quality assessment model, which can not only speed up the training speed, but also reduce the risk of overfitting. At the same time, high-quality feature input enables the network quality assessment model to learn and extract key information more efficiently, improving the overall training and reasoning efficiency.

[0059] 5. Through the high-quality features generated by the feature engineering layer, the network quality assessment model can more accurately assess the quality of the current frequency band and decide whether to switch to another frequency band, which helps to avoid unnecessary frequency band switching, reduce switching delays and packet loss, and thus improve user experience. For example, for video streaming applications, the network quality assessment model can give priority to frequency bands with high throughput and low latency; while for web browsing, it may pay more attention to signal stability.

[0060] The first hidden layer includes a first fully connected layer and a second fully connected layer, and the first fully connected layer and the second fully connected layer are both provided with a ReLU activation function; the second hidden layer includes a third fully connected layer and a fourth fully connected layer, and the third fully connected layer and the fourth fully connected layer are both provided with a ReLU activation function. In addition, the first hidden layer and the second hidden layer both use residual links, that is, the input of the first fully connected layer is jump-connected to the output of the second fully connected layer, and the input of the third fully connected layer is jump-connected to the output of the fourth fully connected layer. The first hidden layer and the second hidden layer take the feature vector output by the feature engineering layer as input, perform linear transformation, and then apply the ReLU activation function to generate an intermediate feature vector; the second hidden layer takes the output of the first hidden layer as input, and finally outputs a high-dimensional feature vector, and the high-dimensional feature vector specifically includes:

[0061] 1. Transformation results of numerical features: For example, data such as signal strength (RSSI), signal-to-noise ratio (SNR), bit error rate (BER), real-time data throughput, average delay, and packet loss rate are converted into new numerical forms after being processed by the hidden layer, reflecting higher-level features. The higher-level features specifically include:

[0062] 1.1 Combination characteristics: For example, there is a certain combination relationship between signal strength (RSSI) and signal-to-noise ratio (SNR). In certain application types (such as video streaming), when RSSI and SNR are both within a certain range, the network quality will be significantly reduced.

[0063] 1.2 Trends and Patterns: The model can gradually extract more complex features through multiple hidden layers, which may represent certain trends or patterns. For example, the model may identify a pattern that when the real-time data throughput gradually decreases and the average latency gradually increases, it indicates that the current frequency band may no longer be suitable for use.

[0064] 1.3 Context-sensitive features: The network quality assessment model can learn features related to a specific context. For example, for a static viewing user behavior pattern, the model may generate a feature to represent the sensitivity to signal strength in this mode. Context-sensitive features enable the model to make more personalized decisions based on different situations, rather than relying solely on a single indicator.

[0065] 2. Encoding of classification features: For example, application types and user behavior patterns are one-hot encoded and further fused with other features through hidden layers to form a more complex representation. For example, assume the following features:

[0066] Application type: [1, 0, 0] indicates video streaming

[0067] User behavior pattern: [0, 1, 0] indicates static viewing

[0068] Signal strength (RSSI): -70 dBm

[0069] Signal-to-Noise Ratio (SNR): 15 dB

[0070] Bit Error Rate (BER): 0.001

[0071] After processing through the hidden layers, a more complex representation like the following example is obtained:

[0072] 2.1 Feature Interaction: The model may learn that “video streaming + static viewing” does not perform well under low signal strength, and thus tends to recommend switching frequency bands under these conditions.

[0073] 2.2 Non-linear Transformation: The model may find that the performance of the video stream drops sharply in a certain RSSI range, while it is relatively stable in another range.

[0074] 2.3 Multi-level abstraction: The first hidden layer may extract features such as "the performance of the video stream under low signal-to-noise ratio", and the second hidden layer may further refine "the performance of the video stream under low signal-to-noise ratio and low RSSI" based on this.

[0075] It should be noted that by using residual connections in the first hidden layer and the second hidden layer, the network quality assessment model can further refine new features while maintaining the original features, so that the final feature representation includes both the original low-level features and the high-level features after multi-layer processing, thereby enhancing the learning ability of the network quality assessment model.

[0076] The one-dimensional convolutional neural network (1D-CNN) takes the output of the second hidden layer as input, and performs a convolution operation on the time series data through a sliding window mechanism to capture local patterns in the input data, wherein the local patterns include short-term fluctuations (for example, the signal strength (RSSI) may show a rapid rise or fall in a short period of time. Such short-term fluctuations may be caused by user movement, obstacle occlusion or other environmental factors. 1D-CNN can capture these short-term fluctuations to help the model understand changes in the current network status), periodic patterns (some network quality indicators may show periodic changes. For example, user activities in certain time periods may cause regular peaks and troughs in data throughput. 1D-CNN can identify these periodic patterns to better predict future network status), spikes and anomalies (Some indicators in the network may occasionally have spikes or outliers. For example, bursty data transmission may lead to instantaneous high throughput, or high bit error rate at a certain moment. 1D-CNN can detect these spikes and anomalies and use them as important features for subsequent analysis), smooth trends (some indicators may show smooth trends in a short period of time, such as gradually decreasing signal strength or gradually increasing delay. 1D-CNN can capture these smooth trends and help the model understand the long-term changes in network quality), multi-indicator combination patterns (sometimes multiple indicators will show a certain pattern at the same time. For example, when the signal strength decreases, the signal-to-noise ratio and throughput may also decrease. 1D-CNN can provide a more comprehensive network quality assessment by learning these multi-indicator combination patterns). For example, assume that the RSSI data for a period of time is as follows:

[0077] Time point t1: -60 dBm

[0078] Time point t2: -62 dBm

[0079] Time point t3: -65 dBm

[0080] Time point t4: -68 dBm

[0081] Time point t5: -70 dBm

[0082] In this example, the 1D-CNN captures the following local patterns:

[0083] Short-term fluctuations: From t1 to t5, the signal strength shows a gradually decreasing trend.

[0084] Spikes and anomalies: If there is a very low value (such as -80 dBm) suddenly between t3 and t4, 1D-CNN can detect this anomaly.

[0085] Smooth trend: During the entire time period, the signal strength shows a smooth downward trend.

[0086] Multi-indicator combination mode: If the changes of SNR and BER are considered simultaneously, 1D-CNN can capture the correlation between these indicators. For example, when RSSI decreases, SNR also decreases, while BER increases.

[0087] By capturing the local patterns described above, 1D-CNN can help the network quality assessment model better understand and process time series data, thereby improving the accuracy of network quality assessment.

[0088] The recurrent neural network (RNN) takes the output of the one-dimensional convolutional neural network (1D-CNN) as input to capture the changing trend of indicators such as signal strength and throughput in the input data over time. The recurrent neural network (RNN) can capture the changing pattern of network quality indicators over time. For example, if the signal strength is found to be gradually decreasing, the RNN can predict that the signal may become weaker in the future, so as to make a decision on frequency band switching in advance.

[0089] Furthermore, by combining 1D-CNN and RNN, the network quality assessment model can more comprehensively consider various patterns and dependencies in time series data, thereby improving the accuracy of network quality assessment. Among them, 1D-CNN can capture local patterns of different scales through convolution kernels of different sizes. For example, small convolution kernels can capture subtle changes, while large convolution kernels can capture wider patterns. In this way, 1D-CNN can provide multi-level feature representation. The recurrent neural network (RNN) can take the multi-scale features extracted by 1D-CNN as input and further process these features. Through the memory mechanism of RNN, it can fuse local features at different time points into multi-scale features, thereby forming a comprehensive understanding of the entire time series. This fusion of multi-scale features helps the model to more comprehensively understand the complex patterns in the data. By combining 1D-CNN and RNN, the network quality assessment model can learn and understand data from multiple perspectives, wherein 1D-CNN provides an understanding of local patterns, while RNN provides an understanding of long-term dependencies. This multi-angle learning approach enables the network quality assessment model to better generalize to unseen data, thereby making more reasonable frequency band switching decisions at different times and in different situations.

[0090] The adaptive weighted fusion layer uses the CQI, ASI and UBP as input features and performs weighted fusion based on the following formula:

[0091]

[0092] in, , and Respectively , and The weight coefficient of Represents a bias term, which is used to adjust the reference value of the final result.

[0093] The adaptive weighted fusion layer generates a single, comprehensive feature vector by fusing the comprehensive quality index CQI, the application sensitivity index ASI and the user behavior pattern UBP. In this way, the network quality assessment model can make comprehensive use of multiple types of feature information instead of relying on only one feature, thereby improving the robustness of the network quality assessment model. Even if some features are abnormal or missing, the network quality assessment model can still rely on other features to make reasonable decisions. For example, if the signal strength (RSSI) data is incomplete or interfered with in a certain period of time, the network quality assessment model can rely more on other features such as signal-to-noise ratio (SNR) and bit error rate (BER) for evaluation. In addition, the adaptive weighted fusion layer can dynamically adjust the feature weights according to the specific needs and behavior patterns of users, thereby providing personalized frequency band switching strategies. For example, for users who often watch video streams, the network quality assessment model can rely more on features such as real-time data throughput and latency; while for users who mainly browse the web, it may rely more on features such as signal stability and packet loss rate.

[0094] In summary, the adaptive weighted fusion layer generates a comprehensive feature representation by adaptively adjusting the weights of different features, thereby improving the accuracy, robustness and generalization ability of network quality assessment, enabling the network quality assessment model to make more reasonable frequency band switching decisions at different times and in different situations, thereby improving user experience.

[0095] The self-attention mechanism layer is used to receive the output of the adaptive fusion layer and highlight key features. The self-attention mechanism layer calculates the attention score by the following formula :

[0096]

[0097] The weighted features are expressed as:

[0098]

[0099] in, and represents the weight matrix, Represents the bias vector.

[0100] The self-attention mechanism layer can calculate the correlation between each feature and other features, so as to dynamically assign weights to each feature. For example, at a certain point in time, if the signal strength (RSSI) is particularly important for the frequency band switching decision, the self-attention mechanism layer will automatically increase the weight of RSSI. By dynamically adjusting the feature weights, the network quality assessment model reduces its dependence on a single feature, thereby improving the robustness of the model. In addition, the self-attention mechanism layer can also dynamically adjust the feature weights according to the specific needs and behavior patterns of the user, thereby providing a personalized frequency band switching strategy. For example, for users who often watch video streams, the self-attention mechanism layer can rely more on features such as real-time data throughput and latency; while for users who mainly browse the web, they may rely more on features such as signal stability and packet loss rate. In summary, the self-attention mechanism layer dynamically adjusts the feature weights, so that the network quality assessment model can make more reasonable frequency band switching decisions at different times and in different situations, thereby improving the user experience.

[0101] The output layer is used to receive the output of the self-attention mechanism layer and output the probability of switching to the 2.4G frequency band. The output layer adopts the Sigmoid activation function to ensure that the output is in the range of [0,1].

[0102] Below, this application collects network data of 5G frequency band and 2.4G frequency band to fully illustrate the method described in this application. The specific examples are as follows:

[0103] Raw data of 5G frequency band

[0104] RSSI: -70 dBm

[0105] SNR: 15 dB

[0106] BER: 0.001

[0107] Real-time data throughput: 300 Mbps

[0108] Average latency: 10 ms

[0109] Packet loss rate: 0.1%

[0110] Application type: Video stream (encoded as [1, 0, 0])

[0111] User behavior pattern: static viewing (encoded as [0, 1, 0])

[0112] Raw data of 2.4G band

[0113] RSSI: -60 dBm

[0114] SNR: 20 dB

[0115] BER: 0.0005

[0116] Real-time data throughput: 100 Mbps

[0117] Average latency: 20 ms

[0118] Packet loss rate: 0.05%

[0119] Application type: Video stream (encoded as [1, 0, 0])

[0120] User behavior pattern: static viewing (encoded as [0, 1, 0])

[0121] After normalizing the above data, we get:

[0122] 5G frequency bands:

[0123] RSSI: 0.55

[0124] SNR: 0.33

[0125] BER: 0.9

[0126] Real-time data throughput: 0.8

[0127] Average latency: 0.1

[0128] Packet loss rate: 0.99

[0129] 2.4G frequency band:

[0130] RSSI: 0.67

[0131] SNR: 0.5

[0132] BER: 0.95

[0133] Real-time data throughput: 0.3

[0134] Average latency: 0.3

[0135] Packet loss rate: 0.995

[0136] The above normalized data of the 5-band and 2.4G bands are respectively input into the network quality assessment model, and the comprehensive quality index (CQI) is first calculated, assuming that the weight of each factor is as follows:

[0137] RSSI: 0.2

[0138] SNR: 0.2

[0139] BER: 0.1

[0140] Real-time data throughput: 0.25

[0141] Average latency: 0.15

[0142] Packet loss rate: 0.1

[0143] For 5G bands:

[0144]

[0145] For 2.4G band:

[0146]

[0147] Secondly, calculate the application sensitivity index (ASI). Among the above indicators, video streaming is more sensitive to real-time data throughput and packet loss rate, so higher weights can be assigned to these two factors. The following is an example:

[0148] Real-time data throughput: 0.7

[0149] Packet loss rate: 0.3

[0150] For 5G bands:

[0151]

[0152] For 2.4G band:

[0153]

[0154] Finally, characterize the user behavior pattern (UBP). For static viewing, we pay more attention to signal strength and bit error rate. We can assign certain weights to these two factors. The example is as follows:

[0155] RSSI: 0.5

[0156] BER: 0.5

[0157] For 5G bands:

[0158]

[0159] For 2.4G band:

[0160]

[0161] After obtaining the CQI, ASI, and UBP of the 5G band and the 2.4G band, the final network quality evaluation result can be obtained through adaptive weighted fusion. Assume that the weights are:

[0162] CQI: 0.4

[0163] ASI: 0.3

[0164] UBP: 0.3

[0165] For 5G bands:

[0166]

[0167] For 2.4G band:

[0168]

[0169] Based on the above calculation process, we can see that the weighted feature score of the 5G band is 0.6926, and the weighted feature score of the 2.4G band is 0.61495. The preset probability threshold for switching to the 2.4G band is 0.8. Since the model predicts that the probability of switching to the 2.4G band is lower than 0.8, the switch from the 5G band to the 2.4G band will not be executed. Staying in the 5G band will be more suitable for the current network usage requirements.

[0170] In another exemplary embodiment, in step S304, training the network quality assessment model includes the following steps:

[0171] S3041: Collect a large amount of real network data including indicators such as signal strength (RSSI), signal-to-noise ratio (SNR), bit error rate (BER), real-time data throughput, average delay, packet loss rate, etc. to form a network data set. This data should cover different time periods and multiple user behavior patterns;

[0172] S3042: Divide the network data set into a training set and a validation set in proportion, for example, the division ratio may be 7:3;

[0173] S3043: Set training parameters, for example, the learning rate is 0.001, the batch size is set to 32, the maximum number of training times is set to 100, the model is trained using the training set, and the loss of the current batch is calculated using the cross entropy loss function until the loss function converges;

[0174] S3044: Use the validation set to validate the trained model, and use precision, recall, F1 score, and mAP as evaluation indicators to validate the performance of the model. When the evaluation indicators all reach 90%, the model is validated. Otherwise, adjust the training parameters (for example, the learning rate is adjusted to 0.01, the batch size is adjusted to 64, and the number of training times is adjusted to 200) or expand the training samples to retrain the model until the model is validated.

[0175] In another exemplary embodiment, in step S3043, the cross entropy loss function is expressed as:

[0176]

[0177] in, represents the weight of each sample, represents the true label, represents the predicted probability, Represents the total number of samples.

[0178] In practical applications, wireless terminals may remain in a certain frequency band (such as the 5G band) in most cases, while switching to another frequency band (such as the 2.4G band) is relatively rare. This class imbalance will cause the model to be more inclined to predict the class with a larger proportion during the training process, that is, not switching the frequency band. Through the weighted cross entropy loss function, a higher weight can be given to the minority class (the case of switching frequency bands), so that more attention can be paid during the training process to improve the recognition ability of the network quality assessment model for the minority class. In addition, by weighting the minority class, the network quality assessment model will pay more attention to samples that are easily overlooked during the training process, thereby improving the prediction accuracy of these samples. In the scenario of wireless terminal frequency band switching, this means that the network quality assessment can more accurately identify wireless terminals that need to switch frequency bands, thereby reducing misjudgments and missed judgments.

[0179] Based on the above embodiments, Figure 3 , for an explanation of the computer-readable storage medium of the exemplary embodiment of the present application, please refer to Figure 3 , the computer-readable storage medium shown is a CD 40, on which a computer program (i.e., a program product) is stored. When the computer program is executed by the processor, the steps recorded in the above method implementation will be implemented, for example, the signal strength of all wireless terminals connected to the 5G frequency band is continuously monitored by the FTTR device; when the 5G signal strength of a target wireless terminal is detected to be lower than a preset threshold, the FTTR device simultaneously sends a BSS Transition Management Request message to the target wireless terminal through the 5G and 2.4G frequency bands; when the target wireless terminal receives at least one BSS Transition Management Request message, the network quality of the 5G and 2.4G frequency bands is evaluated through the FTTR device; the FTTR device determines whether to perform a switching operation on the target wireless terminal based on the network quality evaluation result. The specific implementation method of each step is not repeated here.

[0180] It should be noted that the computer-readable storage medium includes but is not limited to phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical or magnetic storage media, which are not listed here one by one.

[0181] Based on the above embodiments, the present application also provides an electronic device, as shown below: Figure 4 An electronic device for downloading a file according to an exemplary embodiment of the present application is described.

[0182] Figure 4 A block diagram of an exemplary electronic device 50 suitable for implementing the embodiments of the present application is shown, and the electronic device 50 may be a computer system or a cloud server. Figure 4 The electronic device 50 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0183] like Figure 4 As shown, the electronic device 50 includes but is not limited to: one or more processors or processing units 501, a system memory 502, and a bus 503 connecting different system components (including the system memory 502 and the processing unit 501).

[0184] The electronic device 50 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 50, including volatile and non-volatile media, removable and non-removable media.

[0185] The system memory 502 may include computer system readable media in the form of volatile memory, such as RAM 5021 and / or cache memory 5022. The electronic device 50 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the ROM 5023 may be used to read and write non-removable, non-volatile magnetic media ( Figure 4 is not shown in the Figure 4 As shown in the figure, a disk drive for reading and writing a removable non-volatile disk (such as a "floppy disk") and an optical disk drive for reading and writing a removable non-volatile optical disk (such as a CD-ROM, DVD-ROM or other optical media) can be provided. In these cases, each drive can be connected to the bus 503 through one or more data medium interfaces. The system memory 502 may include at least one program product, which has a set (such as at least one) of program modules, which are configured to perform the functions of each embodiment of the present application.

[0186] A program / utility 5025 having a set (at least one) of program modules 5024 may be stored, for example, in system memory 502, and such program modules 5024 include, but are not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. Program modules 5024 generally perform the functions and / or methods of the embodiments described herein.

[0187] The electronic device 50 may also communicate with one or more external devices 504 (e.g., a keyboard, a pointing device, a display, etc.). Such communication may be performed via an I / O interface 505. Furthermore, the electronic device 50 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 506. Figure 4 As shown, the network adapter 506 communicates with other modules (such as the processing unit 501, etc.) of the electronic device 50 via the bus 503. It should be understood that although Figure 4 Not shown, other hardware and / or software modules may be used in conjunction with the electronic device 50 .

[0188] The processing unit 501 executes various functional applications and data processing by running the program stored in the system memory 502, for example, continuously monitoring the signal strength of all wireless terminals connected to the 5G frequency band through the FTTR device; when the 5G signal strength of a target wireless terminal is detected to be lower than a preset threshold, the FTTR device simultaneously sends a BSS Transition Management Request message to the target wireless terminal through the 5G and 2.4G frequency bands; when the target wireless terminal receives at least one BSS Transition Management Request message, the network quality of the 5G and 2.4G frequency bands is evaluated through the FTTR device; the FTTR device determines whether to perform a switching operation on the target wireless terminal based on the network quality evaluation result. The specific implementation method of each step is not repeated here. It should be noted that although several units / modules or sub-units / sub-modules of the file concurrent download device are mentioned in the above detailed description, this division is only exemplary and not mandatory. In fact, according to the implementation of the present application, the features and functions of two or more units / modules described above can be concretized in one unit / module. On the contrary, the features and functions of one unit / module described above may be further divided into multiple units / modules to be embodied.

[0189] In the description of the present application, it should be noted that the terms "first", "second" and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0190] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0191] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0192] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0193] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0194] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that can be executed by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a cloud server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0195] The above embodiments are only for illustrating the technical concept and features of the present application, and their purpose is to enable people familiar with the technology to understand the content of the present application and implement it accordingly, and they cannot be used to limit the protection scope of the present application. Any equivalent changes or modifications made according to the spirit of the present application should be included in the protection scope of the present application.

Claims

1. A method for switching frequency bands of wireless terminals under FTTR equipment, characterized in that: The method comprises: Continuously monitor the signal strength of all wireless terminals connected to the 5G frequency band through FTTR equipment; When the 5G signal strength of a target wireless terminal is detected to be lower than a preset threshold, the FTTR device sends a BSS Transition Management Request message to the target wireless terminal via both the 5G and 2.4G frequency bands; When the target wireless terminal receives at least one BSS Transition Management Request message, the network quality of the 5G and 2.4G frequency bands is evaluated through the FTTR device; The network quality evaluation of the 5G and 2.4G frequency bands includes: Collect network data for both 5G and 2.4G frequency bands, including signal strength, signal-to-noise ratio, bit error rate, real-time data throughput, packet loss rate, average latency, application type, and user behavior patterns; Preprocessing the network data; Build and train a network quality assessment model; Inputting the preprocessed network data into the trained network quality assessment model to predict the probability of switching to the 2.4G frequency band; The network quality assessment model includes: The input layer, feature engineering layer, first hidden layer, second hidden layer, one-dimensional convolutional neural network, recurrent neural network, adaptive weighted fusion layer, self-attention mechanism layer and output layer are connected in sequence; The FTTR device determines whether to perform a handover operation on the target wireless terminal based on the network quality evaluation result.

2. The method for switching frequency band of a wireless terminal under FTTR equipment according to claim 1, characterized in that: If the network quality assessment model predicts that the probability of switching to the 2.4G frequency band is higher than a preset threshold, the switching operation is performed; otherwise, the current 5G frequency band is maintained.

3. The method for switching frequency band of a wireless terminal under FTTR equipment according to claim 1, characterized in that: Preprocessing the network data includes: Normalizing the numerical data in the network data; One-hot encoding categorical features in the network data; Performing missing value processing on the normalized and one-hot encoded network data; The network data after missing value processing is enhanced.

4. The method for switching frequency band of a wireless terminal under FTTR equipment according to claim 1, characterized in that: The first hidden layer includes a first fully connected layer and a second fully connected layer, the second hidden layer includes a third fully connected layer and a fourth fully connected layer, and both the first hidden layer and the second hidden layer use residual connection.

5. The method for switching frequency band of a wireless terminal under FTTR equipment according to claim 1, characterized in that: The network quality assessment model is trained by the following steps: Collect real network data to construct a network dataset; Divide the network dataset into training set and validation set in proportion; Set the training parameters, train the model using the training set, and use the cross entropy loss function to calculate the loss of the current batch until the loss function converges; The trained model is verified using the validation set, and the performance of the model is verified using precision, recall, F1 score, and mAP as evaluation indicators. When all of the evaluation indicators reach 90%, the model is verified. Otherwise, the training parameters are adjusted or the training samples are expanded to retrain the model until the model is verified.

6. A storage medium, characterized in that: The method comprises instructions, which, when executed on a computer, enable the computer to execute the method according to any one of claims 1 to 5.

7. An electronic device, characterized in that: The electronic device comprises: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method according to any one of claims 1 to 5 is implemented.

Images