Method and apparatus for transmitting power management of wireless access points

Abstract

This application discloses a method and apparatus for managing the transmit power of a wireless access point. The method includes: acquiring a real-time load state sequence of a first wireless access point within a first time period; analyzing the real-time load state sequence using a target time series data prediction model to obtain a first predicted load state sequence of the first wireless access point within a second time period, wherein the target time series data prediction model includes multiple memory units corresponding to each time period, and each memory unit includes: a first forget gate, a second forget gate, an input gate, and an output gate, with the output of the first forget gate serving as the input of the second forget gate; and managing the transmit power of the first wireless access point based on the first predicted load state sequence. This application solves the technical problem in related technologies where adjusting the transmit power of a wireless access point does not consider load changes, resulting in poor adaptability of the adjustment results.

Description

Technical Field

[0001] This application relates to the field of device management technology, and more specifically, to a method and apparatus for managing the transmit power of a wireless access point. Background Technology

[0002] With the rapid development of information technology, the construction and management of campus networks face increasing energy consumption and environmental impact issues. Therefore, wireless access points (APs) have become an indispensable part of campus networks to provide flexible and extensive network coverage. Currently, when installing wireless access points, their power is typically set in two ways: Method 1: Static configuration strategy, which sets the power based on experience or estimated peak demand, and the power remains constant throughout the usage period. However, while simple and easy to implement, this power setting method lacks flexibility and cannot adapt to real-time changes in network demand, often leading to energy waste or insufficient network coverage during peak hours; Method 2: Dynamic adjustment strategy, such as time-based scheduling (e.g., reducing power at night) or simple threshold adjustments based on the number of user connections. While this power setting strategy is an improvement over static configuration and can reflect changes in network usage to some extent, it still lacks fine-grained control and predictive capabilities, failing to optimize resource utilization.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This application provides a method and apparatus for managing the transmit power of a wireless access point, which at least solves the technical problem that related technologies do not consider load changes when adjusting the transmit power of a wireless access point, resulting in poor adaptability of the adjustment results.

[0005] According to one aspect of the embodiments of this application, a method for managing the transmit power of a wireless access point is provided, comprising: acquiring a real-time load state sequence of a first wireless access point within a first time period; analyzing the real-time load state sequence using a pre-trained target time series data prediction model to obtain a first predicted load state sequence of the first wireless access point within a second time period, wherein the target time series data prediction model includes multiple memory units corresponding to each time period, each memory unit including: a first forget gate, a second forget gate, an input gate, and an output gate, and the output of the first forget gate is the input of the second forget gate, the second time period being a time period following the first time period; and managing the transmit power of the first wireless access point based on the first predicted load state sequence.

[0006] Optionally, the training process of the target time series data prediction model includes: acquiring multiple sets of training samples, wherein each set of training samples includes: a first load state sequence of the second wireless access point in a first historical time period, and a second load state sequence of the second wireless access point in a second historical time period, wherein the second historical time period is a time period after the first historical time period; constructing an initial time series data prediction model; and iteratively training the initial time series data prediction model using multiple sets of training samples to obtain the target time series data prediction model.

[0007] Optionally, multiple sets of training sample data are acquired, including: identifying multiple second wireless access points; for each second wireless access point, collecting a first initial load state sequence within a first historical time period and a second initial load state sequence within a second historical time period using a network monitoring tool, wherein the first initial load state sequence and the second initial load state sequence include at least one of the following: number of device connections, data throughput, and frequency band usage data; performing preprocessing operations on the first initial load state sequence and the second initial load state sequence to obtain the first load state sequence of the second wireless access point within the first historical time period and the second load state sequence of the second wireless access point within the second historical time period.

[0008] Optionally, the initial time series data prediction model is iteratively trained using multiple sets of training samples to obtain the target time series data prediction model. This includes: traversing each set of training sample data and performing the following steps to obtain the second load state sequence of the second wireless access point within the second historical time period in the training sample data: Step 1: Input the memory unit output corresponding to the previous historical time period of the first historical time period corresponding to the training sample data and the first load state sequence within the training sample data into the first Sigmoid activation function, and multiply the obtained output by a preset short-term time scale factor to obtain the first forget gate activation value; Step 2: Input the memory unit state corresponding to the previous historical time period of the first historical time period corresponding to the training sample data into the first Sigmoid activation function. Multiply by the first forget gate activation value to update the memory unit state corresponding to the previous historical time period of the first historical time period in the training sample data. To obtain the state of the memory cell and the state of the memory cell As the output of the first forget gate; the third step: take the first load state sequence and memory unit state from the training sample data. The first step involves inputting the memory unit output corresponding to the previous historical time period of the first historical time period of the training sample data, along with preset context information, into the first sigmoid activation function. The output is then multiplied by a preset long-term time scale factor to obtain the second forget gate activation value. The fourth step involves using the second forget gate activation value to determine the input gate activation value, and then multiplying the output of the first forget gate by the second forget gate activation value to obtain the memory unit state. , to the state of the memory cell The output of the second forget gate is used as the first candidate state value. The fifth step involves integrating the first load state sequence from the training sample data with the memory unit output from the previous historical time period corresponding to the first historical time period of the training sample data using the tanh activation function to obtain the first candidate state value. The sixth step multiplies the input gate activation value with the first candidate state value to determine the important information to be retained in the first candidate state value, and adds this important information to the output of the second forget gate to obtain the memory unit state corresponding to the first historical time period of the training sample data. Step 7: Use the tanh activation function to process the memory cell state. Processing is performed to obtain the second candidate state value; Step 8: The first load state sequence in the training sample data, the memory unit output and memory unit state corresponding to the previous historical time period of the first historical time period of the training sample data are processed. The input is fed into the second Sigmoid activation function to obtain the output gate activation value; Step 9: Multiply the second candidate state value and the output gate activation value to obtain the memory unit output of the first historical time period corresponding to the training sample data; Step 10: Based on the memory unit output of the first historical time period corresponding to the training sample data, use the initial time series data prediction model to output the corresponding second predicted load state sequence; Construct the target loss function using the second load state sequence in each group of training samples and the corresponding second predicted load state sequence, and adjust the model parameters of the initial time series data prediction model according to the target loss function until the model parameters converge, thus obtaining the trained target time series data prediction model.

[0009] Optionally, managing the transmit power of the first wireless access point based on the first predicted load state sequence includes: determining whether the first predicted load state sequence is empty; disconnecting the power supply to the first wireless access point if the first predicted load state sequence is empty; determining the relationship between the first predicted load state sequence and a preset threshold value if the first predicted load state sequence is not empty; reducing the transmit power of the first wireless access point if the first predicted load state sequence is less than the threshold value; and increasing the transmit power of the first wireless access point if the first predicted load state sequence is not less than the threshold value.

[0010] Optionally, reducing the transmission power of the first wireless access point includes: determining the number of devices using each wireless frequency band based on the type of wireless frequency band used by the multiple terminal devices connected to the first wireless access point, wherein the types of wireless frequency bands include: 5GHz band and 2.4GHz band; reducing the power of the wireless channel corresponding to the 2.4GHz band of the first wireless access point when the number of devices using the 5GHz band is higher than the number of devices using the 2.4GHz band; and reducing the power of the wireless channel corresponding to the 5GHz band of the first wireless access point when the number of devices using the 5GHz band is not higher than the number of devices using the 2.4GHz band.

[0011] Optionally, increasing the transmit power of the first wireless access point includes: increasing the gain of the transmit power amplifier within the first wireless access point; or, adjusting the frequency band information of the first wireless access point, wherein the frequency band information includes at least one of the following: bandwidth and center frequency.

[0012] According to another aspect of the embodiments of this application, a wireless access point transmit power management device is also provided, comprising: an acquisition module, configured to acquire a real-time load state sequence of a first wireless access point within a first time period; a prediction module, configured to analyze the real-time load state sequence using a pre-trained target time series data prediction model to obtain a first predicted load state sequence of the first wireless access point within a second time period, wherein the target time series data prediction model includes multiple memory units corresponding to each time period, each memory unit including: a first forget gate, a second forget gate, an input gate, and an output gate, and the output of the first forget gate is the input of the second forget gate, and the second time period is a time period following the first time period; and a management module, configured to manage the transmit power of the first wireless access point according to the first predicted load state sequence.

[0013] According to another aspect of the embodiments of this application, a computer program product is also provided, the computer program product comprising: a computer program, wherein the computer program, when executed by a processor, implements the above-described wireless access point transmit power management method.

[0014] According to another aspect of the embodiments of this application, an electronic device is also provided, the electronic device including: a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the above-described wireless access point transmit power management method through the computer program.

[0015] In this embodiment, the real-time load state sequence of the first wireless access point during a first time period is analyzed using a target time series data prediction model to obtain a first predicted load state sequence of the first wireless access point during a second time period. Based on this first predicted load state sequence, the transmit power of the first wireless access point is dynamically and intelligently managed. This prediction-based power management strategy achieves the technical effect of accurately analyzing the predicted load state of the first wireless access point during the second time period, achieving the goal of dynamically optimizing the power of the wireless access point. This effectively saves energy and improves network service quality, thereby solving the technical problem that related technologies do not consider load changes when adjusting the transmit power of wireless access points, resulting in poor adaptability of the adjustment results. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0017] Figure 1 This is a flowchart illustrating an optional wireless access point transmit power management method according to an embodiment of this application;

[0018] Figure 2 This is a schematic diagram of the structure of an optional initial time series data prediction model according to an embodiment of this application;

[0019] Figure 3 This is a schematic diagram of the structure of an optional wireless access point transmit power management device according to an embodiment of this application;

[0020] Figure 4 This is a schematic diagram of the structure of an optional electronic device according to an embodiment of this application. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0022] It should be noted that the terms "first," "second," etc., used in the specification, claims, and drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] To better understand the embodiments of this application, the following is a translation and explanation of some nouns or terms that appear in the description of the embodiments of this application:

[0024] A wireless access point (AP) is an access point for a wireless network, commonly known as a "hotspot." Access points mainly come in two forms: integrated routing and switching access devices and pure access point devices. Integrated devices handle access and routing power and are generally the core of the wireless network; while pure access devices only handle wireless client access and are typically used as wireless network extensions, connecting with other APs or the main AP to expand wireless coverage.

[0025] LSTM (Long Short-Term Memory) network: This is an improved RNN (Recurrent Neural Network) designed to solve the gradient vanishing and gradient exploding problems that occur during long sequence training of RNNs. LSTM controls the flow of information through gating mechanisms, including forget gates, input gates, and output gates, where:

[0026] (1) Input Gate: This controls how much of the current input information is updated into the memory unit. It receives the current input vector and the hidden state vector from the previous time step as input, and calculates an update ratio between 0 and 1 using the Sigmoid activation function. Therefore, the expression for the input gate can be written as:

[0027]

[0028] In the formula This represents the Sigmoid activation function of the input gate. This represents the weight matrix of the input gate; This represents the bias vector of the input gate. This represents the input vector at the current time t. This represents the hidden state vector at the previous time step t-1.

[0029] (2) Forget Gate: This gate determines which information in the memory unit should be retained and which should be discarded. It takes the current input vector and the hidden state vector from the previous time step as input, and maps them to values ​​between 0 and 1 using a sigmoid activation function. Values ​​close to 0 indicate that the corresponding information will be forgotten, and values ​​close to 1 indicate that the corresponding information will be retained. Therefore, the expression for the forget gate can be written as:

[0030]

[0031] In the formula The sigmoid activation function represents the forget gate. The weight matrix represents the forget gate. This represents the bias vector of the forget gate.

[0032] (3) Candidate Cell State: The current input vector and the hidden state vector from the previous time step are transformed using the tanh activation function, and then the two are multiplied to obtain the information that needs to be updated in the memory cell. Therefore, the expression for the candidate memory cell can be written as:

[0033]

[0034] In the formula The weight matrix represents the candidate memory units. This represents the bias vector of the candidate memory cell.

[0035] (4) Update the cell state: Update the cell state based on the results of the forget gate and the input gate. Therefore, the expression for updating the cell state can be written as:

[0036]

[0037] In the formula This represents the state of the memory cell at the current time t. This indicates the state of the memory cell at the previous time step t-1. This indicates element-wise multiplication.

[0038] (5) Output Gate: This gate determines which information from the memory cell will be output as the current hidden state. It receives the current input vector and the hidden state vector from the previous time step as input, calculates an output ratio between 0 and 1 using the Sigmoid activation function, and then multiplies it by the memory cell processed by the tanh activation function to obtain the current hidden state. Therefore, the expression for the output gate can be written as:

[0039]

[0040]

[0041] In the formula This represents the weight matrix of the output gate; This represents the bias vector of the output gate.

[0042] Mean Squared Error (MSE) is a measure of the difference between observed and true values. It is the average of the sum of squared errors and is used to evaluate the accuracy of a predictive model. A smaller MSE indicates a higher accuracy in describing the experimental data. Therefore, the expression for MSE can be written as:

[0043]

[0044] In the formula, n represents the total number of samples. This represents the i-th observation. This represents the i-th true value.

[0045] Root Mean Squared Error (RMSE): This is the square root of the mean squared error, measuring the deviation between the observed value and the true value. Therefore, the expression for the mean squared error can be written as:

[0046]

[0047] In the formula, n represents the total number of samples. This represents the i-th observation. This represents the i-th true value.

[0048] Mean Absolute Error (MAE): This is the average of absolute errors; it measures the average error between the observed and the true values. Therefore, the expression for mean absolute error can be written as:

[0049]

[0050] In the formula, n represents the total number of samples. This represents the i-th observation. This represents the i-th true value.

[0051] Example 1

[0052] According to an embodiment of this application, a method for managing the transmit power of a wireless access point is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0053] Figure 1 This is a flowchart illustrating a method for managing the transmit power of a wireless access point according to an embodiment of this application. Figure 1 As shown, the method includes the following steps:

[0054] Step S102: Obtain the real-time load status sequence of the first wireless access point within the first time period.

[0055] In the technical solution provided in step S102 above, the real-time load status sequence can be understood as the usage data of the first wireless access point in the current time period, which provides intuitive feedback on network usage patterns.

[0056] Step S104: Analyze the real-time load state sequence using a pre-trained target time series data prediction model to obtain the first predicted load state sequence of the first wireless access point in the second time period.

[0057] In the technical solution provided in step S104 above, the target time series data prediction model has an independent memory unit (Cell State) in each time period. Each memory unit includes a first forget gate, a second forget gate, an input gate, and an output gate. The output of the first forget gate is the input of the second forget gate, thus enabling it to effectively handle long-term dependencies in the sequence. When processing time series data, the model can learn the long-term trend of information and "recall" states from long ago when needed. That is, the memory unit corresponding to each time period interacts with its input and the state of the previous memory unit, forming an information flow that runs through the entire time series to retain and transmit long-term dependency information. This ensures that the target time series data prediction model accurately predicts the future load state sequence of the wireless access point in the second time period (i.e., a future time period after the first time period).

[0058] Step S106: Manage the transmit power of the first wireless access point according to the first predicted load state sequence.

[0059] In the technical solution provided in step S106 above, the first predicted load state sequence output by the target time series data prediction model is used as the decision basis to dynamically adjust the transmission power of the first wireless access point, thereby enabling the first wireless access point to efficiently respond to the dynamic changes in network demand and achieve effective energy use.

[0060] Based on the scheme defined in steps S102 to S106 above, it can be understood that in this embodiment, the real-time load state sequence of the first wireless access point in the first time period is analyzed by the target time series data prediction model to obtain the first predicted load state sequence of the first wireless access point in the second time period. Based on this first predicted load state sequence, the transmit power of the first wireless access point is dynamically and intelligently managed. This prediction-based power management strategy achieves the technical effect of accurately analyzing the predicted load state of the first wireless access point in the second time period, achieving the goal of dynamically optimizing the power of the wireless access point, thereby effectively saving energy and improving network service quality.

[0061] The following section describes each step of the wireless access point transmit power management method in conjunction with a specific implementation process.

[0062] As an optional implementation, the training process of the above-mentioned target time series data prediction model may include:

[0063] Step S11: Obtain multiple sets of training samples. Each set of training samples includes: a first load state sequence of the second wireless access point within a first historical time period, and a second load state sequence of the second wireless access point within a second historical time period, wherein the second historical time period is a time period following the first historical time period.

[0064] Step S12: Construct an initial time series data prediction model. This initial time series data prediction model is based on a Long Short-Term Memory (LSTM) network, which includes multiple memory units corresponding to different time periods. Each memory unit includes a first forget gate, a second forget gate, an input gate, and an output gate. The output of the first forget gate is the input of the second forget gate, as shown below. Figure 2 As shown.

[0065] Step S13: Iteratively train the initial time series data prediction model using multiple sets of training samples to obtain the target time series data prediction model.

[0066] Optionally, in the technical solution provided in step S11 above, the method includes:

[0067] Step S111: Determine multiple second wireless access points. These second wireless access points may be wireless access points located within the same network coverage area as the first wireless access point.

[0068] Step S112, for each second wireless access point, repeat the following steps:

[0069] First, network monitoring tools are used to collect the first initial load state sequence of the second wireless access point within the first historical time period and the second initial load state sequence within the second historical time period. The first and second initial load state sequences include, but are not limited to, the number of device connections (i.e., the number of terminals connected to the second wireless access point), data throughput (also known as data traffic, i.e., the amount of uploaded data and downloaded data), frequency band usage data (i.e., the number of connected devices and the amount of data transmitted per frequency band), signal strength, etc. Then, the first and second initial load state sequences are preprocessed to obtain the first load state sequence of the second wireless access point within the first historical time period and the second load state sequence of the second wireless access point within the second historical time period. The preprocessing operations include at least one of the following: data cleaning (removing invalid or abnormal data points), data formatting (unifying the data format), feature engineering (standardizing, normalizing, or extracting higher-order statistical features from the data), etc.

[0070] Optionally, in the technical solution provided in step S13 above, the method includes:

[0071] Step S131: Traverse each group of training sample data and perform the following steps to obtain the second load state sequence of the second wireless access point within the second historical time period in the training sample data:

[0072] Step 1: Input the memory unit output corresponding to the previous time period of the first historical time period in the training sample data and the first load state sequence in the training sample data into the first Sigmoid activation function, and multiply the resulting output by a preset short-term time scale factor to obtain the first forget gate activation value. Therefore, the expression for the first forget gate activation value can be written as:

[0073]

[0074] in, This represents the first Sigmoid activation function of the first forget gate. This represents the weight matrix of the first Sigmoid activation function. Let represent the bias vector of the first Sigmoid activation function, and The parameters can be continuously updated during model training. Additionally, This represents the first load state sequence corresponding to the first historical time period. This represents the hidden state vector of the time period preceding the first historical time period. Represents the short-term time scale factor, and , It is an adjustable parameter used to control the rate of change of short-term time scale factors, where a larger value... It can make the model forget short-term information more quickly, while smaller This allows the model to retain more short-term information; in the embodiments of this application, it is preferred to... Set it to a larger value, such as 0.8-1.2.

[0075] Step 2: Enter the memory unit state corresponding to the previous historical time period of the first historical time period in the training sample data. Multiply by the first forget gate activation value to update the memory unit state corresponding to the previous time period of the first historical time period in the training sample data. To obtain the state of the memory cell and the state of the memory cell As the output of the first forget gate. Therefore, the expression for the output of the first forget gate can be written as:

[0076]

[0077] Step 3: Extract the first load state sequence and memory cell states from the training sample data. The memory unit output corresponding to the previous time period of the first historical time period of the training sample data, along with preset context information, is output to the first Sigmoid activation function. The resulting output is then multiplied by a preset long-term time scale factor to obtain the second forgetting gate activation value. Therefore, the expression for the second forgetting gate activation value can be written as:

[0078]

[0079] in, Let represent the first sigmoid activation function of the second forget gate, and the weight matrix and bias vector of the activation functions of the first and second forget gates are the same. This indicates the context information of the second wireless access point within the first historical time period, including but not limited to: timestamp information (such as hour, date, day of the week, etc.), seasonal information (different seasons or holidays), access point location information, network configuration information (such as wireless LAN configuration, tunnel server configuration), special event information (such as large-scale events, conferences, etc.), and weather information. Additionally, Represents the long-term time scale factor, and , It is an adjustable parameter used to control the rate of change of long-term time scale factors, where a larger value... It can make the model forget long-term information more quickly, while smaller This allows the model to retain more long-term information; in the embodiments of this application, it is preferred to... Set it to a smaller value, such as 0.1-0.3.

[0080] Step 4: Determine the input gate activation value based on the second forget gate activation value. Therefore, the expression for the input gate activation value can be written as:

[0081]

[0082] Simultaneously, the output of the first forget gate is multiplied by the activation value of the second forget gate to obtain the memory cell state. and the state of the memory cell As the output of the second forget gate. Therefore, the expression for the output of the second forget gate can be written as:

[0083]

[0084] Step 5: Integrate the first load state sequence from the training sample data with the memory unit output corresponding to the previous time period of the first time period in the training sample data using the tanh activation function to obtain the first candidate state value. Therefore, the expression for the first candidate state value can be written as:

[0085]

[0086] in, This represents the tanh activation function. This represents the weight matrix of the tanh activation function. Let represent the bias vector of the tanh activation function, and The parameters can be continuously updated during model training.

[0087] Step 6: Multiply the input gate activation value by the first candidate state value to determine the important information to be retained in the first candidate state value, and add the important information to the output of the second forget gate to obtain the memory unit state of the first historical time period corresponding to the training sample data. Therefore, the state of the memory cell The expression can be written as:

[0088]

[0089] Step 7: Use the tanh activation function to process the memory cell state. After processing, the second candidate state value is obtained. Therefore, the expression for the second candidate state value can be written as:

[0090]

[0091] Step 8: Combine the first load state sequence in the training sample data, the memory unit output and memory unit state corresponding to the previous time period of the first historical time period in the training sample data. The input is fed into the second Sigmoid activation function to obtain the output gate activation value. Therefore, the expression for the output gate activation value can be written as:

[0092]

[0093] in, This represents the second Sigmoid activation function of the output gate. This represents the weight matrix of the second Sigmoid activation function. Let represent the bias vector of the second Sigmoid activation function, and The parameters can be continuously updated during model training.

[0094] Step 9: Multiply the second candidate state value and the output gate activation value to obtain the memory unit output for the first historical time period corresponding to the training sample data. Therefore, the expression for the memory unit output for the first historical time period can be written as:

[0095]

[0096] Step 10: Based on the memory unit output of the first time period corresponding to the training sample data, use the initial time series data to predict the output of the model for the corresponding second predicted load state sequence.

[0097] In the above process, the second load state sequence is input into the model, so that the load state sequence is forgotten once and then forgotten again. The forget gate activation value obtained from the second forgetting allows the model to learn the coupling relationship between time series variables in depth, thereby making fuller use of the historical information to be forgotten and improving the accuracy of prediction.

[0098] Step S132: Construct a target loss function using the second load state sequence and the corresponding second predicted load state sequence in each group of training sample data, and adjust the model parameters of the initial time series data prediction model according to the target loss function until the model parameters converge, thus obtaining the trained target time series data prediction model.

[0099] The target loss function can be selected from one of the following: mean squared error (MSE), root mean square error (RMSE), or mean absolute error (MAE). Furthermore, the model parameters include, but are not limited to: the weight matrix and bias vector of the sigmoid activation function and the tanh activation function, the learning rate, the batch size, and the dropout rate.

[0100] Furthermore, after completing the training through the above steps S11-S13, a target time series data prediction model that has been trained can be obtained. The model is then used to analyze the real-time load state sequence of the first wireless access point in the first time period to obtain the first predicted load state sequence of the first wireless access point in the second time period.

[0101] Furthermore, in the technical solution provided in step S106 above, the transmit power of the first wireless access point can be managed based on the first predicted load state sequence and according to the following strategy:

[0102] First, determine whether the first predicted load state sequence is empty, that is, determine whether the first wireless access point has no load in the second time period.

[0103] If the first predicted load state sequence is empty, it indicates whether the first wireless access point has no load in the second time period. At this time, a shutdown command can be issued from the switch side to disconnect the power supply of the first wireless access point in order to save energy.

[0104] If the first predicted load state sequence is not empty, it indicates that the first wireless access point has a load in the second time period. At this time, the relationship between the first predicted load state sequence and the preset threshold value can be further determined to judge the load size.

[0105] If the first predicted load state sequence is less than the threshold value, it indicates that the load of the first wireless access point is relatively small in the second time period. At this time, the transmission power of the first wireless access point can be appropriately reduced.

[0106] If the first predicted load state sequence is not less than the threshold value, it indicates that the load of the first wireless access point is relatively large in the second time period. At this time, the transmission power of the first wireless access point can be appropriately increased.

[0107] Optionally, when the load on the first wireless access point is low during the second time period, the transmit power of the first wireless access point can be reduced using the following strategies:

[0108] Based on the type of wireless frequency band used by each of the multiple terminal devices connected to the first wireless access point, the number of devices using each type of wireless frequency band is determined. The types of wireless frequency bands include: 5GHz band and 2.4GHz band.

[0109] The fact that the number of devices using the 5GHz band is higher than the number of devices using the 2.4GHz band indicates that during periods of low load, most terminal devices support and prefer to use the 5GHz band, thus allowing for a reduction in the power of the wireless channel corresponding to the 2.4GHz band at the first wireless access point.

[0110] If the number of devices using the 5GHz band is no higher than the number of devices using the 2.4GHz band, it indicates that during periods of low load, most terminal devices support and prefer to use the 2.4GHz band. Therefore, the power of the wireless channel corresponding to the 5GHz band of the first wireless access point can be reduced.

[0111] This approach ensures a good user experience while reducing interference and improving network efficiency. Besides the power reduction methods listed above, those skilled in the art can also reduce the power of wireless access points through other technical solutions. For example, switching frequency bands based on interference detection results during low load periods or switching frequency bands based on signal quality results during low load periods should also be within the scope of this invention.

[0112] Optionally, when the load on the first wireless access point is high during the second time period, the transmit power of the first wireless access point can be increased by optimizing the radio frequency parameters of the first wireless access point, including: increasing the gain of the transmit power amplifier in the first wireless access point; or, adjusting the frequency band information of the first wireless access point, wherein the frequency band information includes at least one of the following: bandwidth and center frequency.

[0113] Specifically, optimizing the radio frequency parameters of the first wireless access point typically involves adjusting the wireless transmission characteristics of the access point to improve signal quality and coverage. This includes, but is not limited to: adjusting the gain of connected antennas (i.e., adjusting the ability of the antenna connected to the wireless access point to amplify signals in a specific direction), adjusting the directivity of connected antennas (i.e., adjusting the radiation patterns of the antenna connected to the wireless access point in different directions), adjusting the frequency band information of the wireless access point (i.e., adjusting the characteristics of the wireless access point in a specific frequency band, such as bandwidth and center frequency), and adjusting the performance parameters of the transmit power amplifier within the wireless access point (including but not limited to power output, gain, linearity, efficiency, and noise figure). Similarly, in addition to the power-increasing schemes listed above, those skilled in the art can also reduce the power of the wireless access point through other technical solutions. For example, configuring the antennas (including gain and directivity) according to the deployment environment of the wireless access point, or increasing the power of the wireless access point according to the number of terminals in the coverage area of ​​the wireless access point, should also be within the scope of protection of this invention.

[0114] Based on the wireless access point transmit power management method described above, it is easy to see that the embodiments of this application have the following technical advantages compared with traditional wireless access point power adjustment schemes:

[0115] (1) The target time series data prediction model based on the improved long short-term memory network is used to analyze the real-time load state sequence. On the one hand, it can capture long-term dependencies from historical data and improve prediction accuracy. On the other hand, after the real-time load state sequence has gone through two forgetting processes, it can fully select the information to be forgotten and further improve the accuracy of the prediction results.

[0116] (2) The first predicted load state sequence of the first wireless access point in the future time period is accurately predicted by the target time series data prediction model. Compared with simple threshold adjustment or time-based power management, the embodiments of this application can make forward-looking predictions and responses in advance.

[0117] (3) The transmit power of the first wireless access point is intelligently managed based on the first predicted load state sequence to ensure that the transmit power of the first wireless access point is reduced during the predicted low load period to reduce energy consumption; while during the predicted high load period, the transmit power of the first wireless access point is adjusted to a higher level in advance to meet user needs and avoid network congestion. This ensures both network performance and effective energy saving.

[0118] Example 2

[0119] According to an embodiment of this application, a wireless access point transmit power management device for implementing the wireless access point transmit power management method in Embodiment 1 is also provided, such as... Figure 3 As shown, the transmit power management device of the wireless access point includes at least: an acquisition module 32, a prediction module 34, and a management module 36, wherein:

[0120] The acquisition module 32 is used to acquire the real-time load status sequence of the first wireless access point within a first time period.

[0121] The prediction module 34 is used to analyze the real-time load state sequence using a pre-trained target time series data prediction model to obtain the first predicted load state sequence of the first wireless access point in the second time period.

[0122] The target time series data prediction model includes multiple memory units corresponding to different time periods. Each memory unit includes a first forget gate, a second forget gate, an input gate, and an output gate. The output of the first forget gate is the input of the second forget gate, and the second time period is a time period following the first time period.

[0123] The management module 36 is used to manage the transmit power of the first wireless access point based on the first predicted load state sequence.

[0124] In addition, the wireless access point's transmit power management device also includes a model training module, which is used to pre-train the target time series data prediction model.

[0125] As an optional implementation, the above-mentioned model training module can train the target time series data prediction model according to the following process:

[0126] Step S21: Obtain multiple sets of training samples.

[0127] Each training sample includes: a first load state sequence of the second wireless access point within a first historical time period, and a second load state sequence of the second wireless access point within a second historical time period, wherein the second historical time period is a time period following the first historical time period.

[0128] Step S22: Construct an initial time series data prediction model.

[0129] The aforementioned initial time series data prediction model is constructed based on a long short-term memory network, which includes multiple memory units corresponding to different time periods. Each memory unit includes a first forget gate, a second forget gate, an input gate, and an output gate, with the output of the first forget gate serving as the input of the second forget gate.

[0130] Step S23: Iteratively train the initial time series data prediction model using multiple sets of training samples to obtain the target time series data prediction model.

[0131] Optionally, in the technical solution provided in step S21 above, the method includes:

[0132] Step S211: Determine multiple second wireless access points. These second wireless access points may be wireless access points located within the same network coverage area as the first wireless access point.

[0133] Step S212, for each second wireless access point, repeat the following steps:

[0134] First, network monitoring tools are used to collect the first initial load state sequence of the second wireless access point within the first historical time period and the second initial load state sequence within the second historical time period. The first and second initial load state sequences include, but are not limited to, the number of device connections (i.e., the number of terminals connected to the second wireless access point), data throughput (also known as data traffic, i.e., the amount of uploaded data and downloaded data), frequency band usage data (i.e., the number of connected devices and the amount of data transmitted per frequency band), signal strength, etc. Then, the first and second initial load state sequences are preprocessed to obtain the first load state sequence of the second wireless access point within the first historical time period and the second load state sequence of the second wireless access point within the second historical time period. The preprocessing operations include at least one of the following: data cleaning (removing invalid or abnormal data points), data formatting (unifying the data format), feature engineering (standardizing, normalizing, or extracting higher-order statistical features from the data), etc.

[0135] Optionally, in the technical solution provided in step S23 above, the method includes:

[0136] Step S231: Traverse each group of training sample data and perform the following steps to obtain the second load state sequence of the second wireless access point within the second historical time period in the training sample data:

[0137] Step 1: Input the memory unit output corresponding to the previous time period of the first historical time period in the training sample data and the first load state sequence in the training sample data into the first Sigmoid activation function, and multiply the resulting output by a preset short-term time scale factor to obtain the first forget gate activation value. Therefore, the expression for the first forget gate activation value can be written as:

[0138]

[0139] in, This represents the first Sigmoid activation function of the first forget gate. This represents the weight matrix of the first Sigmoid activation function. Let represent the bias vector of the first Sigmoid activation function, and The parameters can be continuously updated during model training. Additionally, This represents the first load state sequence corresponding to the first historical time period. This represents the hidden state vector of the time period preceding the first historical time period. Represents the short-term time scale factor, and , It is an adjustable parameter used to control the rate of change of short-term time scale factors, where a larger value... It can make the model forget short-term information more quickly, while smaller This allows the model to retain more short-term information; in the embodiments of this application, it is preferred to... Set it to a larger value, such as 0.8-1.2.

[0140] Step 2: Enter the memory unit state corresponding to the previous historical time period of the first historical time period in the training sample data. Multiply by the first forget gate activation value to update the memory unit state corresponding to the previous time period of the first historical time period in the training sample data. To obtain the state of the memory cell and the state of the memory cell As the output of the first forget gate. Therefore, the expression for the output of the first forget gate can be written as:

[0141]

[0142] Step 3: Extract the first load state sequence and memory cell states from the training sample data. The memory unit output corresponding to the previous time period of the first historical time period of the training sample data, along with preset context information, is output to the first Sigmoid activation function. The resulting output is then multiplied by a preset long-term time scale factor to obtain the second forgetting gate activation value. Therefore, the expression for the second forgetting gate activation value can be written as:

[0143]

[0144] in, Let represent the first sigmoid activation function of the second forget gate, and the weight matrix and bias vector of the activation functions of the first and second forget gates are the same. This indicates the context information of the second wireless access point within the first historical time period, including but not limited to: timestamp information (such as hour, date, day of the week, etc.), seasonal information (different seasons or holidays), access point location information, network configuration information (such as wireless LAN configuration, tunnel server configuration), special event information (such as large-scale events, conferences, etc.), and weather information. Additionally, Represents the long-term time scale factor, and , It is an adjustable parameter used to control the rate of change of long-term time scale factors, where a larger value... It can make the model forget long-term information more quickly, while smaller This allows the model to retain more long-term information; in the embodiments of this application, it is preferred to... Set it to a smaller value, such as 0.1-0.3.

[0145] Step 4: Determine the input gate activation value based on the second forget gate activation value. Therefore, the expression for the input gate activation value can be written as:

[0146]

[0147] Simultaneously, the output of the first forget gate is multiplied by the activation value of the second forget gate to obtain the memory cell state. and the state of the memory cell As the output of the second forget gate. Therefore, the expression for the output of the second forget gate can be written as:

[0148]

[0149] Step 5: Integrate the first load state sequence from the training sample data with the memory unit output corresponding to the previous time period of the first time period in the training sample data using the tanh activation function to obtain the first candidate state value. Therefore, the expression for the first candidate state value can be written as:

[0150]

[0151] in, This represents the tanh activation function. This represents the weight matrix of the tanh activation function. Let represent the bias vector of the tanh activation function, and The parameters can be continuously updated during model training.

[0152] Step 6: Multiply the input gate activation value by the first candidate state value to determine the important information to be retained in the first candidate state value, and add the important information to the output of the second forget gate to obtain the memory unit state of the first historical time period corresponding to the training sample data. Therefore, the state of the memory cell The expression can be written as:

[0153]

[0154] Step 7: Use the tanh activation function to process the memory cell state. After processing, the second candidate state value is obtained. Therefore, the expression for the second candidate state value can be written as:

[0155]

[0156] Step 8: Combine the first load state sequence in the training sample data, the memory unit output and memory unit state corresponding to the previous time period of the first historical time period in the training sample data. The input is fed into the second Sigmoid activation function to obtain the output gate activation value. Therefore, the expression for the output gate activation value can be written as:

[0157]

[0158] in, This represents the second Sigmoid activation function of the output gate. This represents the weight matrix of the second Sigmoid activation function. Let represent the bias vector of the second Sigmoid activation function, and The parameters can be continuously updated during model training.

[0159] Step 9: Multiply the second candidate state value and the output gate activation value to obtain the memory unit output for the first historical time period corresponding to the training sample data. Therefore, the expression for the memory unit output for the first historical time period can be written as:

[0160]

[0161] Step 10: Based on the memory unit output of the first time period corresponding to the training sample data, use the initial time series data to predict the output of the model for the corresponding second predicted load state sequence.

[0162] In the above process, the second load state sequence is input into the model, so that the load state sequence is forgotten once and then forgotten again. The forget gate activation value obtained from the second forgetting allows the model to learn the coupling relationship between time series variables in depth, thereby making fuller use of the historical information to be forgotten and improving the accuracy of prediction.

[0163] Step S232: Construct a target loss function using the second load state sequence and the corresponding second predicted load state sequence in each training sample, and adjust the model parameters of the initial time series data prediction model according to the target loss function until the model parameters converge, thus obtaining the trained target time series data prediction model.

[0164] Furthermore, after the model training module completes training through the aforementioned steps S21-S23, a trained target time series data prediction model can be obtained. Then, the prediction module 34 can call this model to analyze the real-time load state sequence of the first wireless access point within the first time period, and obtain the first predicted load state sequence of the first wireless access point within the second time period.

[0165] Optionally, the management module 36 can manage the transmit power of the first wireless access point based on the first predicted load state sequence and according to the following strategy:

[0166] First, determine whether the first predicted load state sequence is empty, that is, determine whether the first wireless access point has no load in the second time period.

[0167] If the first predicted load state sequence is empty, it indicates whether the first wireless access point has no load in the second time period. At this time, a shutdown command can be issued from the switch side to disconnect the power supply of the first wireless access point in order to save energy.

[0168] If the first predicted load state sequence is not empty, it indicates that the first wireless access point has a load in the second time period. At this time, the relationship between the first predicted load state sequence and the preset threshold value can be further determined to judge the load size.

[0169] If the first predicted load state sequence is less than the threshold value, it indicates that the load of the first wireless access point is relatively small in the second time period. At this time, the transmission power of the first wireless access point can be appropriately reduced.

[0170] If the first predicted load state sequence is not less than the threshold value, it indicates that the load of the first wireless access point is relatively large in the second time period. At this time, the transmission power of the first wireless access point can be appropriately increased.

[0171] Optionally, when the load on the first wireless access point is low during the second time period, the management module 36 may reduce the transmit power of the first wireless access point using the following strategies:

[0172] Based on the type of wireless frequency band used by each of the multiple terminal devices connected to the first wireless access point, the number of devices using each type of wireless frequency band is determined. The types of wireless frequency bands include: 5GHz band and 2.4GHz band.

[0173] The fact that the number of devices using the 5GHz band is higher than the number of devices using the 2.4GHz band indicates that during periods of low load, most terminal devices support and prefer to use the 5GHz band, thus allowing for a reduction in the power of the wireless channel corresponding to the 2.4GHz band at the first wireless access point.

[0174] If the number of devices using the 5GHz band is no higher than the number of devices using the 2.4GHz band, it indicates that during periods of low load, most terminal devices support and prefer to use the 2.4GHz band. Therefore, the power of the wireless channel corresponding to the 5GHz band of the first wireless access point can be reduced.

[0175] This approach ensures a good user experience while reducing interference and improving network efficiency. Besides the power reduction methods listed above, those skilled in the art can also reduce the power of wireless access points through other technical solutions. For example, switching frequency bands based on interference detection results during low load periods or switching frequency bands based on signal quality results during low load periods should also be within the scope of this invention.

[0176] Optionally, when the load on the first wireless access point is high during the second time period, the management module 36 can optimize the radio frequency parameters of the first wireless access point to increase the transmit power of the first wireless access point, including: increasing the gain of the transmit power amplifier in the first wireless access point; or, adjusting the frequency band information of the first wireless access point, wherein the frequency band information includes at least one of the following: bandwidth and center frequency.

[0177] Specifically, optimizing the radio frequency parameters of the first wireless access point typically involves adjusting the wireless transmission characteristics of the access point to improve signal quality and coverage. This includes, but is not limited to: adjusting the gain of connected antennas (i.e., adjusting the ability of the antenna connected to the wireless access point to amplify signals in a specific direction), adjusting the directivity of connected antennas (i.e., adjusting the radiation patterns of the antenna connected to the wireless access point in different directions), adjusting the frequency band information of the wireless access point (i.e., adjusting the characteristics of the wireless access point in a specific frequency band, such as bandwidth and center frequency), and adjusting the performance parameters of the transmit power amplifier within the wireless access point (including but not limited to power output, gain, linearity, efficiency, and noise figure). Similarly, in addition to the power-increasing schemes listed above, those skilled in the art can also reduce the power of the wireless access point through other technical solutions. For example, configuring the antennas (including gain and directivity) according to the deployment environment of the wireless access point, or increasing the power of the wireless access point according to the number of terminals in the coverage area of ​​the wireless access point, should also be within the scope of protection of this invention.

[0178] It should be noted that each module in the wireless access point transmit power management device in this application embodiment corresponds one-to-one with each implementation step of the wireless access point transmit power management method in embodiment 1. Since embodiment 1 has been described in detail, some details not shown in this embodiment can be referred to embodiment 1, and will not be elaborated further here.

[0179] Example 3

[0180] According to an embodiment of this application, a computer program product is also provided, which includes a computer program, wherein when the computer program is executed by a processor, it implements the wireless access point transmit power management method in Embodiment 1.

[0181] According to an embodiment of this application, a non-volatile storage medium is also provided, which includes a stored computer program, wherein the device containing the non-volatile storage medium executes the wireless access point transmit power management method in Embodiment 1 by running the computer program.

[0182] According to an embodiment of this application, a processor is also provided for running a computer program, wherein the computer program executes the wireless access point transmit power management method in Embodiment 1.

[0183] According to an embodiment of this application, an electronic device is also provided, comprising: a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the wireless access point transmit power management method of Embodiment 1 through the computer program.

[0184] Specifically, the computer program executes the following steps during runtime: acquiring the real-time load state sequence of the first wireless access point within a first time period; analyzing the real-time load state sequence using a pre-trained target time series data prediction model to obtain the first predicted load state sequence of the first wireless access point within a second time period, wherein the target time series data prediction model includes multiple memory units corresponding to each time period, and each memory unit includes: a first forget gate, a second forget gate, an input gate, and an output gate, and the output of the first forget gate is the input of the second forget gate, and the second time period is a time period following the first time period; and managing the transmission power of the first wireless access point based on the first predicted load state sequence.

[0185] As an alternative implementation, the above-mentioned electronic device may exist in the form of a mobile terminal, a computer terminal, or a similar computing device. Figure 4 A hardware block diagram of an electronic device for implementing a transmit power management method for a wireless access point is shown. Figure 4 As shown, the electronic device 40 may include one or more (shown as 402a, 402b, ..., 402n) processors 402 (processors 402 may include, but are not limited to, microprocessors such as MCUs or programmable logic devices such as FPGAs), a memory 404 for storing data, and a transmission device 406 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 4 The structure shown is for illustrative purposes only and does not limit the structure of the electronic device described above. For example, electronic device 40 may also include... Figure 4More or fewer components than shown, or with Figure 4 Different configurations shown.

[0186] It should be noted that the aforementioned one or more processors 402 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element of the electronic device 40. As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0187] The memory 404 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the wireless access point transmit power management method in this embodiment. The processor 402 executes various functional applications and data processing by running the software programs and modules stored in the memory 404, thereby implementing the above-mentioned application vulnerability detection method. The memory 404 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 404 may further include memory remotely located relative to the processor 402, and these remote memories can be connected to the electronic device 40 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0188] The transmission device 406 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the electronic device 40. In one example, the transmission device 406 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 406 may be a Radio Frequency (RF) module used for wireless communication with the Internet.

[0189] The display can be, for example, a touchscreen liquid crystal display (LCD), which allows the user to interact with the user interface of the electronic device x0.

[0190] The sequence numbers of the above embodiments are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0191] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0192] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0193] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0194] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0195] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0196] The above are merely preferred embodiments of this application. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for managing the transmit power of a wireless access point, characterized in that, include: Obtain the real-time load status sequence of the first wireless access point within the first time period; The real-time load state sequence is analyzed using a pre-trained target time series data prediction model to obtain the first predicted load state sequence of the first wireless access point within a second time period. The target time series data prediction model includes multiple memory units corresponding to different time periods. Each memory unit includes a first forget gate, a second forget gate, an input gate, and an output gate. The output of the first forget gate is the input of the second forget gate. The second time period is a time period following the first time period. The target time series data prediction model is obtained by iteratively training an initial time series data prediction model using multiple sets of training sample data. The first forget gate activation value is obtained by multiplying the output of the memory unit corresponding to the previous historical time period of the first historical time period corresponding to the training sample data by the first sigmoid activation function and the first load state sequence in the training sample data by a preset short-term time scale factor. The second forget gate activation value is obtained by multiplying the output of the first load state sequence in the training sample data, the output of the first forget gate, the memory unit output corresponding to the previous historical time period of the first historical time period corresponding to the training sample data, and preset context information by a preset long-term time scale factor. The transmit power of the first wireless access point is managed based on the first predicted load state sequence.

2. The method according to claim 1, characterized in that, The training process of the target time series data prediction model includes: Multiple sets of training sample data are acquired, wherein each set of training sample data includes: a first load state sequence of the second wireless access point in a first historical time period, and a second load state sequence of the second wireless access point in a second historical time period, wherein the second historical time period is a time period after the first historical time period. Construct an initial time series data prediction model; The initial time series data prediction model is iteratively trained using the multiple sets of training sample data to obtain the target time series data prediction model.

3. The method according to claim 2, characterized in that, Obtain multiple sets of training sample data, including: Identify multiple second wireless access points; For each second wireless access point, a first initial load state sequence and a second initial load state sequence within the first historical time period are collected using network monitoring tools. The first initial load state sequence and the second initial load state sequence include at least one of the following: number of device connections, data throughput, and frequency band usage data. The first initial load state sequence and the second initial load state sequence are preprocessed to obtain the first load state sequence of the second wireless access point within the first historical time period and the second load state sequence of the second wireless access point within the second historical time period.

4. The method according to claim 2, characterized in that, The initial time series data prediction model is iteratively trained using the multiple sets of training sample data to obtain the target time series data prediction model, including... By iterating through each group of training sample data, the following steps are performed to obtain the second load state sequence of the second wireless access point within the second historical time period in the training sample data: Step 1: Input the memory unit output corresponding to the previous historical time period corresponding to the first historical time period of the training sample data and the first load state sequence in the training sample data into the first Sigmoid activation function, and multiply the obtained output result by the preset short-term time scale factor to obtain the first forget gate activation value; Step 2: Multiply the memory unit state corresponding to the previous historical time period of the first historical time period corresponding to the training sample data with the first forget gate activation value, update the memory unit state corresponding to the previous historical time period of the first historical time period corresponding to the training sample data, obtain the memory unit state, and use the memory unit state as the output of the first forget gate. Step 3: Input the first load state sequence in the training sample data, the memory unit state, the memory unit output corresponding to the previous historical time period of the first historical time period corresponding to the training sample data, and the preset context information into the first Sigmoid activation function, and multiply the obtained output result by the preset long-term time scale factor to obtain the second forget gate activation value. Step 4: Use the second forget gate activation value to determine the input gate activation value, and multiply the output of the first forget gate by the second forget gate activation value to obtain the memory cell state. Use the memory cell state as the output of the second forget gate. Step 5: Use the tanh activation function to integrate the first load state sequence in the training sample data and the memory unit output corresponding to the previous historical time period of the first historical time period of the training sample data to obtain the first candidate state value; Step 6: Multiply the input gate activation value with the first candidate state value to determine the important information to be retained in the first candidate state value, and add the important information to the output of the second forget gate to obtain the memory unit state of the first historical time period corresponding to the training sample data; Step 7: Process the state of the memory cell using the tanh activation function to obtain the second candidate state value; Step 8: Input the first load state sequence in the training sample data, the memory unit output corresponding to the previous historical time period of the first historical time period corresponding to the training sample data, and the memory unit state into the second Sigmoid activation function to obtain the output gate activation value; Step 9: Multiply the second candidate state value and the output gate activation value to obtain the memory unit output of the first historical time period corresponding to the training sample data; Step 10: Based on the memory unit output of the first historical time period corresponding to the training sample data, use the initial time series data prediction model to output the corresponding second predicted load state sequence; A target loss function is constructed using the second load state sequence and the corresponding second predicted load state sequence within each set of training sample data. The model parameters of the initial time series data prediction model are then adjusted according to the target loss function until the model parameters converge, thus obtaining the trained target time series data prediction model.

5. The method according to claim 1, wherein Managing the transmit power of the first wireless access point based on the first predicted load state sequence includes: Determine whether the first predicted load state sequence is empty; If the first predicted load state sequence is empty, disconnect the power supply to the first wireless access point; If the first predicted load state sequence is not empty, determine the relationship between the first predicted load state sequence and the preset threshold value. If the first predicted load state sequence is less than the threshold value, reduce the transmit power of the first wireless access point; If the first predicted load state sequence is not less than the threshold value, increase the transmit power of the first wireless access point.

6. The method according to claim 5, characterized in that, Reducing the transmit power of the first wireless access point includes: The number of devices using each wireless frequency band is determined based on the type of wireless frequency band used by the multiple terminal devices connected to the first wireless access point, wherein the types of wireless frequency bands include: 5GHz band and 2.4GHz band; If the number of devices using the 5GHz band is higher than the number of devices using the 2.4GHz band, reduce the power of the wireless channel corresponding to the 2.4GHz band of the first wireless access point; If the number of devices using the 5GHz band is not higher than the number of devices using the 2.4GHz band, reduce the power of the wireless channel corresponding to the 5GHz band of the first wireless access point.

7. The method according to claim 1, characterized in that, Increasing the transmit power of the first wireless access point includes: Increase the gain of the transmit power amplifier within the first wireless access point; or adjust the frequency band information of the first wireless access point, wherein the frequency band information includes at least one of the following: bandwidth and center frequency.

8. A wireless access point transmit power management device, characterized in that, include: The acquisition module is used to acquire the real-time load status sequence of the first wireless access point within a first time period. The prediction module is used to analyze the real-time load state sequence using a pre-trained target time series data prediction model to obtain a first predicted load state sequence of the first wireless access point within a second time period. The target time series data prediction model includes multiple memory units corresponding to different time periods. Each memory unit includes a first forget gate, a second forget gate, an input gate, and an output gate. The output of the first forget gate is the input of the second forget gate. The second time period is a time period following the first time period. The target time series data prediction model is obtained by iteratively training an initial time series data prediction model using multiple sets of training sample data. The first forget gate activation value is obtained by multiplying the output of the memory unit corresponding to the previous historical time period of the first historical time period corresponding to the training sample data by the first sigmoid activation function and the first load state sequence in the training sample data by a preset short-term time scale factor. The second forget gate activation value is obtained by multiplying the output of the first load state sequence in the training sample data, the output of the first forget gate, the memory unit output corresponding to the previous historical time period of the first historical time period corresponding to the training sample data, and preset context information by a preset long-term time scale factor. The management module is used to manage the transmit power of the first wireless access point based on the first predicted load state sequence.

9. A computer program product, characterized in that, include: A computer program, wherein when executed by a processor, the computer program implements the transmit power management method for a wireless access point as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: A memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the transmit power management method for a wireless access point according to any one of claims 1 to 7 via the computer program.

Images