Wireless local area network throughput prediction method and storage medium

Abstract

The invention relates to a wireless local area network throughput prediction method and a storage medium. The wireless local area network throughput prediction method comprises the following steps: acquiring network topology, signal strength, a channel access mechanism, MCS and spatial fluxion combination, PER, test duration and AP effective transmission sequence data; and inputting the initial physical model to calculate the initial throughput. And performing linear regression fitting on the initial prediction and actual measurement throughput to obtain a slope correction coefficient and an intercept correction term which are used for constructing a correction model. The model is cross-validated and trained. And inputting the network topology, the RSSI and the channel access mechanism to predict the overall throughput of the network. The dynamic correction model solves the problem of prediction deviation caused by neglecting transient fluctuation of a wireless channel.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication network optimization, and in particular to a wireless local area network throughput prediction method and storage medium. Background Art

[0002] With the evolution of the Wi-Fi 7 standard and the widespread deployment of high-density access points (APs), co-channel interference and data collisions caused by overlapping signal coverage between adjacent APs have become increasingly prominent. In dense AP scenarios, competition for channel resources intensifies, and dynamic physical layer factors (such as the modulation and coding scheme (MCS) and the number of spatial streams (NSS)) are coupled with link quality indicators (such as RSSI), resulting in a significant decrease in system throughput. Traditional methods rely on large-scale signal strength (RSSI) and static network topology modeling, but are unable to capture the transient fluctuations of wireless channels (such as micro-scale fading and burst interference). This results in a significant deviation between throughput predictions and actual performance, limiting network optimization effectiveness.

[0003] Current mainstream solutions fall into two categories: data-driven models (such as XGBoost, a gradient boosting decision tree) and physical analysis models. The former predicts throughput by inputting parameters such as RSSI, PHYRate, and MCS / NSS. However, in a real-world scenario with two APs, the mean square error (MSE) of the predictions was as high as 137.51 (RMSE = 11.73). This error is primarily due to RSSI's partiality for large-scale information. The latter calculates throughput based on the following theoretical formula:

[0004]

[0005] However, because they fail to incorporate dynamic channel uncertainty, the predicted values ​​are generally overstated, reaching an MSE of 169.62, even inferior to data-driven models. Both approaches struggle to balance accuracy and real-time performance.

[0006] The rapidly time-varying characteristics of wireless channels (such as transient interference and fluctuations in packet loss rates (PERs)) invalidate the theoretical assumptions of physical models. Purely data-driven models lack physical constraints and therefore lack generalization capabilities. In high-density AP scenarios (e.g., ≥3 APs), system complexity surges, with the MSE rising to 248.01. The prediction accuracy (0.86–0.88) is significantly lower than in low-density scenarios (0.91–0.92). Existing technologies lack a dynamic correction mechanism for physical models, making it impossible to adaptively adjust model parameters to match actual network conditions, limiting the reliability of throughput predictions. Summary of the Invention

[0007] The present invention provides a wireless local area network throughput prediction method and storage medium, which aims to solve the problem of systematic prediction deviation caused by the physical model not adapting to the dynamic characteristics of the wireless channel in the existing WLAN throughput prediction.

[0008] To achieve the above object, the present invention provides a first aspect of a wireless local area network throughput prediction method, comprising the following steps:

[0009] Obtain network topology data, signal strength indicator data, channel access mechanism data, modulation and coding scheme and spatial stream combination data, packet error rate data, total test duration data, and predicted access point effective transmission time series data;

[0010] Substitute the acquired modulation and coding scheme combination data, spatial stream number data, packet error rate data, total test duration data, and access point effective transmission time series data into the initial physical model to calculate the initial throughput prediction value;

[0011] Using the linear regression algorithm, the calculated initial throughput prediction value is fitted with the measured throughput to obtain the slope correction coefficient and intercept correction term;

[0012] Applying the obtained slope correction coefficient and intercept correction term to the initial physical model to construct a corrected physical model;

[0013] Use cross-validation method to train and verify the constructed modified physical model;

[0014] Based on the modified physical model after training and verification, the obtained network topology data, signal strength indicator data, and channel access mechanism data are input to predict the overall network throughput.

[0015] Furthermore, the method for predicting effective transmission time series data of the access point includes:

[0016] Obtaining the network topology data of the target wireless local area network and extracting the location identification information of the access points and stations therefrom;

[0017] Collecting service flow data of the wireless local area network and encoding the transport layer protocol type into discrete feature values;

[0018] Measuring received signal strength indicator values ​​between all nodes in the network topology data and calculating received signal strength indicator mean characteristics of all communication link pairs;

[0019] Reading a network allocation vector threshold value configured in the wireless local area network and encoding it into a binary feature;

[0020] Combining the extracted location identification information, the encoded discrete feature value, the calculated received signal strength indication mean feature, and the encoded binary feature to form an input feature vector;

[0021] Set the initial hyperparameter search range for the gradient boosted decision tree regression model;

[0022] A grid search algorithm is used to traverse all combinations within the set initial hyperparameter search range, and the root mean square error corresponding to each combination is calculated through five-fold cross validation. The hyperparameter combination with the smallest root mean square error is selected as the optimal hyperparameter combination.

[0023] The optimal hyperparameter combination is used to configure a gradient boosting decision tree regression model, and the model is trained using a training dataset containing input feature vectors and corresponding access point effective transmission time series data;

[0024] The input feature vector constructed by the scenario to be predicted is input into the trained gradient boosting decision tree regression model, and the access point effective transmission time series data is output.

[0025] Furthermore, the method of encoding the transport layer protocol type into a discrete characteristic value includes:

[0026] When the protocol type is User Datagram Protocol, the value is 0;

[0027] When the protocol type is Transmission Control Protocol, the value is 1.

[0028] Furthermore, the five-fold cross validation is performed as follows:

[0029] The training dataset is randomly divided into five mutually exclusive subsets;

[0030] Select one subset as the validation set and the remaining four subsets as the training set;

[0031] Train the model under each hyperparameter combination and calculate the root mean square error on the validation set;

[0032] After five cycles, the average root mean square error is taken as the evaluation index.

[0033] Furthermore, the method for predicting the combined data of the modulation and coding scheme and the number of spatial streams includes:

[0034] Obtaining received signal strength indication data and preset threshold information between at least two access points and a station;

[0035] Determining the synchronous / asynchronous communication state between the access points according to the dynamic relationship between the received signal strength indication data and the preset threshold information;

[0036] Based on the synchronous / asynchronous communication state, classify the signal sent by the adjacent access point to the station as an interference signal, and classify the signal sent by the adjacent station as environmental noise;

[0037] Calculate the signal-to-noise ratio of the site based on the signal classification results;

[0038] Inputting the signal-to-noise ratio, synchronous / asynchronous communication state and access point transmit power into a gradient boosting decision tree model;

[0039] The gradient boosting decision tree model is used to output the prediction results of the modulation and coding scheme of the target access point and the spatial stream number combination data.

[0040] Furthermore, the method for determining the synchronous / asynchronous communication state between access points includes:

[0041] For the downlink data signal of each access point, extract the maximum value and average value of its received signal strength indicator data;

[0042] When the maximum values ​​of the received signal strength indicator data of the downlink data signal of the first access point and the downlink data signal of the second access point do not reach the packet detection threshold and the average values ​​of the received signal strength indicator data do not reach the network allocation vector threshold, it is determined that the two access points are in a synchronous communication state;

[0043] When the maximum value of the received signal strength indicator data of only one of the downlink data signal of the first access point and the downlink data signal of the second access point reaches the packet detection threshold or the average value of the received signal strength indicator reaches the network allocation vector threshold, it is determined that the two access points are in an asynchronous communication state;

[0044] When the maximum values ​​of the received signal strength indication data of the downlink data signal of the first access point and the downlink data signal of the second access point both reach the packet detection threshold or the average values ​​of the received signal strength indication data both reach the network allocation vector threshold, it is determined that the two access points are in a mixed communication state.

[0045] Furthermore, the signal classification rules are:

[0046] Interference signals are defined as signals from non-associated access points;

[0047] The ambient noise floor is defined as the signal from non-target sites.

[0048] Furthermore, the signal-to-noise ratio of the site is calculated:

[0049] When APs are synchronized, the signal-to-noise ratio calculation ignores the ambient noise floor intensity.

[0050] When APs are in an asynchronous state, the signal-to-noise ratio calculation ignores the interference signal strength;

[0051] When APs are in a mixed synchronous and asynchronous state, this value aggregates the linear power of the ambient noise floor and the interference signal.

[0052] Furthermore, the method of aggregating the linear power values ​​of the ambient noise floor and the interference signal includes:

[0053] Convert all signal strength values ​​to be aggregated from dBm to mW.

[0054] Sum the converted mW unit values;

[0055] Convert the summation result back to dBm units.

[0056] To achieve the above object, the second aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the wireless local area network throughput prediction method are executed.

[0057] Beneficial effects of the present invention:

[0058] Compared with existing technologies, the wireless local area network throughput prediction method and storage medium provided by the present invention effectively solves the problem of systematic prediction bias caused by traditional physical models ignoring transient fluctuations in wireless channels by constructing a dynamically corrected physical model. Specifically, a slope correction coefficient and an intercept correction term are introduced on the basis of the traditional physical model, and linear regression is used to dynamically compensate for channel efficiency overestimation (such as transient interference loss) and systematic offset (such as protocol overhead). At the same time, correction parameters are differentiated for different AP density scenarios, and measured physical layer dynamic parameters (modulation coding scheme and spatial stream combination data, packet error rate data) and access point effective transmission time series data are integrated to improve adaptability while retaining physical interpretability. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments.

[0060] Figure 1 This is a flow chart of a wireless local area network throughput prediction method disclosed in an embodiment of the present invention.

[0061] Figure 2 This is a comparison chart of the predicted value and the actual value of a throughput physical model disclosed in an embodiment of the present invention.

[0062] Figure 3 This is a comparison chart of the predicted value and the actual value of a throughput physical model disclosed in an embodiment of the present invention.

[0063] Figure 4 This is a model residual graph and a QQ graph for cases where the number of APs is 2 and the number of APs is 3, disclosed in an embodiment of the present invention.

[0064] Figure 5 This is a distribution characteristic diagram of a cumulative distribution function of throughput disclosed in an embodiment of the present invention.

[0065] Figure 6 It is a complete directed graph of interactions among nodes in a WLAN network disclosed in an embodiment of the present invention.

[0066] Figure 7 This is a schematic diagram of how the root mean square error changes with the increase in training rounds, as disclosed in an embodiment of the present invention.

[0067] Figure 8 This is a diagram of a 2AP sending downlink data to an associated STA disclosed in an embodiment of the present invention.

[0068] Figure 9 This is a flowchart of determining synchronous / asynchronous states when using 2APs, as disclosed in an embodiment of the present invention.

[0069] Figure 10 This is a classification diagram of a 3AP interference signal and environmental background noise disclosed in an embodiment of the present invention.

[0070] Figure 11 This is a flowchart of determining synchronous / asynchronous states of 3APs disclosed in an embodiment of the present invention.

[0071] Figure 12 This is a synchronous / asynchronous state diagram between two APs disclosed in an embodiment of the present invention.

[0072] Figure 13 This is a classification diagram of interference signals and environmental noise floor when there are two APs, as disclosed in an embodiment of the present invention.

[0073] Figure 14 This is a synchronous / asynchronous state diagram between two APs in a number of three disclosed in an embodiment of the present invention.

[0074] Figure 15 This is a diagram of a calculation formula for modifying the signal-to-noise ratio when the number of APs is 2, disclosed in an embodiment of the present invention.

[0075] Figure 16 This is a schematic diagram of a gradient boosting tree training disclosed in an embodiment of the present invention.

[0076] Figure 17 This is a prediction accuracy analysis diagram for a case where the number of APs is 2, as disclosed in an embodiment of the present invention.

[0077] Figure 18 This is a graph of the accuracy of MCS and NSS prediction when the number of APs is 2, disclosed in an embodiment of the present invention.

[0078] Figure 19 This is an importance ranking diagram when the number of APs is 2, disclosed in an embodiment of the present invention.

[0079] Figure 20This is a prediction accuracy diagram when the number of APs is 3, as disclosed in an embodiment of the present invention.

[0080] Figure 21 This is a graph of the accuracy prediction of MCS and NSS when the number of APs is 3, as disclosed in an embodiment of the present invention.

[0081] Figure 22 This is an importance ranking diagram when the number of APs is 3, as disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0082] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described 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 making creative efforts should fall within the scope of protection of the present invention.

[0083] According to an embodiment of the present invention, it should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the following production method, in some cases, the steps shown or described can be executed in an order different from that herein.

[0084] like Figure 1 As shown, the present invention provides a wireless local area network throughput prediction method, comprising the following steps:

[0085] Step S100: Acquire network topology data, signal strength indicator data, channel access mechanism data, modulation and coding scheme and spatial stream combination data, packet error rate data, total test duration data, and predicted access point effective transmission time series data;

[0086] This step involves seven types of data: measured data, configured data, and predicted data. Together, these data constitute the input feature set for the throughput prediction model. The following section explains in detail the definition, source, importance, and role of each data item in the model.

[0087] Network topology data refers to the physical location distribution and connectivity of access points (APs) and stations in a wireless LAN. The network topology determines the physical placement of APs, directly impacting signal propagation paths, coverage, and interference patterns. For example, in a high-density AP deployment, topology data can identify the distances and overlaps between APs, helping to quantify signal conflicts and interference. This data, such as AP location information and station association information, acquired through WLAN management systems or field measurement tools, serves as a fundamental input for throughput prediction and is used to analyze the macro-impact of network structure.

[0088] Signal strength indicator data, also known as received signal strength indicator (RSSI), measures the wireless signal strength between all nodes (such as AP-AP or AP-station). RSSI reflects the quality of the wireless link and has a direct and significant impact on resource allocation and interference assessment. Measured RSSI data represents large-scale channel information but cannot fully capture instantaneous fluctuations. RSSI is used as a key feature in model input to help assess signal attenuation and link stability. This data is collected in real time using wireless sniffers or AP logs.

[0089] Channel access mechanism data refers to the protocol rules and parameters of the WLAN's media access control (MAC) layer, such as the Network Allocation Vector (NAV) threshold or the configuration of the Carrier Sense Multiple Access / Collision Avoidance (CSMA / CA) mechanism. The channel access mechanism is a key factor affecting throughput, as it determines the timing of data frame transmission and the collision handling strategy. For example, the NAV threshold (nav) is encoded as a binary feature to determine channel occupancy. This data, extracted from the AP configuration file, helps the model assess the impact of channel contention and access delay on throughput.

[0090] The combination of the modulation and coding scheme and the number of spatial streams (MCS and NSS) defines the physical layer transmission parameters, including the modulation type, coding rate, and number of spatial streams. Together, they determine the theoretical maximum transmission rate (PHY rate). MCS and NSS are key physical layer factors that directly impact transmission efficiency. In the physical model, the PHY rate is derived from MCS and NSS. These data are obtained through prediction model output during field measurements (see the following section for detailed methods).

[0091] The packet error rate (PER) indicates the probability of errors during data transmission, calculated as the ratio of erroneous packets to total transmitted packets. (1-PER) is defined as the probability of successful transmission. This is a core variable in the physical model and directly influences throughput calculations. PER data is measured using WLAN monitoring tools and reflects the effects of channel noise, interference, and bit error rate. In the model, PER helps adjust the theoretical rate to match the actual environment.

[0092] The total test duration (Test Duration) refers to the total time of the entire WLAN performance test, recorded in seconds. This duration is used as the denominator in the physical throughput model to convert the effective transmission time (Seq_time) and PHY rate into average throughput. This data is captured using test logs or a timer to ensure that throughput predictions are based on a standardized time base, avoiding errors caused by inconsistent time scales.

[0093] The predicted access point effective transmission time series data (Seq_time) refers to the sequence of time segments used by the AP for data transmission, representing the actual duration of the AP's channel occupation during the test. Seq_time is a key input to the physical model, but measured data is often missing. Therefore, it must be pre-derived using a prediction model (such as the gradient boosting decision tree algorithm) (detailed methods for obtaining this data are described below). Seq_time prediction is based on characteristics such as network topology, RSSI mean, and protocol type. Obtaining this data ensures that the model accurately reflects the AP's transmission opportunities.

[0094] Step S200: Substitute the acquired modulation and coding scheme combined data with the number of spatial streams, data packet error rate data, total test duration data, and access point effective transmission time series data into the initial physical model to calculate an initial throughput prediction value;

[0095] Substitute the acquired key data into the initial physical model to calculate the initial throughput prediction value for each access point (AP). This step directly models the transmission characteristics of the physical layer, but does not yet consider the dynamic interference and system deviation in the actual environment. The calculation formula is:

[0096]

[0097] Where: T AP The PHY_Rate is the theoretical maximum transmission rate of the WLAN device at the physical layer, determined by the (MCS, NSS) selected by the AP. The PER is the probability of an error during data packet transmission, and (1-PER) is the probability of successful transmission. The Seq_time is the effective time segment used by the AP for data transmission. The TestDuration is the total duration of the entire test process, in seconds.

[0098] Since Seq_time is missing in the test data set, the prediction of the AP's effective transmission time is used as the input of the above formula. At the same time, the (MCS, NSS), packet loss rate PER, and test data TestDuration in the test data set are substituted into the formula to calculate the throughput of each AP.

[0099] from Figure 2 As can be seen, the throughput predicted by the physical model is slightly higher than the actual throughput. The mean square error (MSE) of the throughput physical model is 169.62. Therefore, the prediction results directly using the throughput physical model are not accurate enough, and are even slightly worse than the prediction results of the gradient boosting decision tree algorithm (MSE = 137.51).

[0100] Step S300: Using a linear regression algorithm, the calculated initial throughput prediction value is fitted with the measured throughput to obtain a slope correction coefficient and an intercept correction term;

[0101] Because wireless channels are ever-changing and real WLAN networks have many uncertainties, the throughput predicted by the physical formula deviates from the throughput measured in the actual environment. Therefore, a regression equation is established:

[0102]

[0103] Among them, α is the linear correction coefficient and β is the intercept coefficient.

[0104] The optimal α and β were obtained by linear fitting (for example, when the number of APs is 2, α = 0.9612, β = -3.7984; when the number of APs is 3, α = 0.9300, β = -6.3337).

[0105] Step S400: Apply the obtained slope correction coefficient and intercept correction term to the initial physical model to construct a corrected physical model;

[0106] The revised throughput calculation formula is:

[0107]

[0108] Step S500: using a cross-validation method to train and validate the constructed modified physical model;

[0109] The training dataset was grouped by the number of APs (e.g., two groups of 2 and 3 APs), with each group randomly divided into five mutually exclusive subsets. In each iteration, one subset was selected as the validation set, and the remaining four subsets were used as the training set. The predicted throughput was calculated using the modified physical model formula and compared with the measured values ​​to calculate the mean squared error (MSE). After five iterations, the average MSE of all validation sets was summarized (e.g., the modified model MSE = 75.43 for 2 APs). The residual distribution and error accumulation function were analyzed to ensure that the model met homoscedasticity and that the residuals were close to a normal distribution.

[0110] Step S600: Based on the modified physical model after training and verification, the obtained network topology data, signal strength indication data, and channel access mechanism data are input to predict the overall network throughput.

[0111] Substitute network topology data, signal strength indicator data, and channel access mechanism data into the modified physical model formula, calculate the throughput of each AP, and then aggregate them into the total network throughput.

[0112] like Figure 3 As shown in the figure, for the case where the number of APs is 2, the corrected MSE value is 75.43, which is about 55.5% lower than the uncorrected MSE value of 169.62; for the case where the number of APs is 3, the corrected MSE value is 68.62, which is about 72.33% lower than the uncorrected MSE value of 248.01. Figure 3 As can be seen from the figure, the prediction effect of the adjusted model is better than that of the unadjusted model. This shows that using linear regression to adjust the physical prediction model is extremely effective, and the prediction results of the tuned physical model are far better than those of the gradient boosting decision tree algorithm.

[0113] In order to verify the fitting effect of the model, a residual graph can be drawn for analysis. The horizontal axis represents the predicted value of the modified model, and the vertical axis represents the residual (actual throughput minus predicted throughput). Figure 4 (a) and (c) show the residual graphs when the number of APs is 2 and the number of APs is 3, respectively.

[0114] At the same time, we draw a QQ graph to test whether the residuals conform to the normal distribution. The horizontal axis is the quantile of the standard normal distribution, and the vertical axis is the quantile of the residual. Figure 4 (b) and (d) show the QQ graphs when the number of APs is 2 and the number of APs is 3, respectively. Figure 4 The residual plots of (a) and (c) show that the residuals are randomly scattered and have no obvious trend or pattern, indicating that the modified physical model fits well and the residuals meet the homoscedasticity assumption. Figure 4 From the QQ plots (b) and (d), we can see that the data points are roughly distributed along the 45-degree line, indicating that the residuals are close to a normal distribution, which meets the normality assumption.

[0115] In order to more comprehensively understand the data distribution characteristics of system throughput and evaluate the accuracy of the model, the error cumulative distribution function (CDF) is introduced for analysis. The error CDF shows the probability that the error is less than or equal to a certain value, which helps to intuitively observe the trends and characteristics of data distribution, such as Figure 5 shown.

[0116] Figure 5 (a) and (c) show the CDF curves of the error when there are 2 APs and 3 APs. Figure 5(b) and (d) show the QQ plots for 2 and 3 APs, respectively. As can be seen from the figures, the CDF of the error gradually increases with increasing throughput, indicating the cumulative probability of error within this range. Furthermore, the QQ plots show that the error distribution is close to a normal distribution, which conforms to the normality assumption.

[0117] In the above step S100, the method for predicting the effective transmission time series data of the access point includes:

[0118] Step S101: Acquire network topology data of a target wireless local area network and extract location identification information of access points and sites therefrom;

[0119] The network topology data is collected in real time through the network management interface of the target wireless LAN (such as SNMP protocol, NetConf configuration interface or dedicated probe equipment). The data includes three major elements: node type and quantity, connection relationship and location attribute: First, the number of physical devices of access points (AP) and stations (STA) and their role identification are clarified. For example, in a 2-AP scenario, a K4 topology containing 2AP+2STA (a total of 4 nodes) is formed, and in a 3-AP scenario, a K6 topology containing 3AP+3STA (a total of 6 nodes) is formed; secondly, the bidirectional communication links between all nodes are recorded to form a complete directed graph structure (K n ),like Figure 6 As shown, each node has a signal interaction path with other nodes. The link direction represents the signal transmission direction (such as AP→STA uplink, STA→STA interference link). Finally, the relative position identification information between the nodes is extracted (denoted as loc_id).

[0120] In the process of extracting location identification information, a unique identifier is first assigned to each AP and an independent identifier is assigned to each STA. The relative position matrix is ​​calculated based on the physical deployment coordinates of the nodes to generate the topological location attribute loc_id. For example, in the K6 topology, loc_id can be mapped to a 6×6 matrix whose element value represents the distance level between two nodes. Secondly, the network topology is abstracted into a weighted directed graph K n =(V,E,ω), where the vertex set V is all AP / STA nodes, the edge set E contains all directed edges, and the weight ω is later used to carry the RSSI value. The loc_id extracted in this step serves as a mathematical representation of the network topology and is directly input into the prediction model to characterize the impact of spatial layout on signal interference (for example, increased interference in high-density areas due to overlapping locations). This feature, together with service traffic, RSSI, and threshold, constitutes the input vector, effectively solving the nonlinear prediction problem of transmission time caused by dynamic topology changes in multi-AP environments.

[0121] Step S102: collecting service flow data of the wireless local area network and encoding the transport layer protocol type into a discrete feature value;

[0122] By deploying the traffic mirroring port on the target wireless LAN core switch or the monitoring interface of the wireless controller, the original business traffic data packets from the network layer to the transport layer are collected in real time, and the transport layer protocol type is identified using deep packet parsing technology: when the UDP protocol is detected, the protocol feature value is encoded as a discrete value of 0 based on its connectionless and low-latency characteristics; when the TCP protocol is detected, the feature value is encoded as a discrete value of 1 based on its connection-oriented and flow control mechanism. This encoding operation is completed in the data preprocessing stage, and ultimately forms a discrete feature sequence with a length equal to the number of traffic samples. Its technical necessity lies in the fact that the transport layer protocol type significantly affects the effective transmission time through the retransmission mechanism and congestion control strategy, and discrete coding can quantify the protocol differences into numerical features, enabling the gradient boosting decision tree model to learn the nonlinear relationship between protocol type and throughput.

[0123] Step S103: measuring the received signal strength indicator values ​​between all nodes in the network topology data, and calculating the received signal strength indicator mean characteristics of all communication link pairs;

[0124] Spectrum analyzers deployed at each node in the wireless LAN periodically measure the received signal strength indicator (RSSI) between all nodes in the network topology. This includes three types of links: interference signal strength between APs, service link strength between STAs and APs, and co-channel interference strength between STAs and STAs. For each link type, three statistical values ​​(sum, max, and mean) are collected according to the requirements in Table 1 (for the 2-AP scenario) and Table 2 (for the 3-AP scenario). Based on the raw RSSI data, mean features are calculated for the bidirectional links (AP0→STA1 and STA1→AP0) between each pair of communicating nodes (for example, AP0→STA1). This generates an RSSI mean feature vector for each communication link pair.

[0125] Table 1: RSSI information between nodes when the number of APs is 2

[0126]

[0127] Table 2: RSSI information between nodes when the number of APs is 3

[0128]

[0129] Step S104: Read the network allocation vector threshold value configured in the wireless local area network and encode it into a binary feature;

[0130] Thresholds include the packet detection threshold (PD), energy detection threshold (ED), and NAV threshold. The PD and ED thresholds detect the maximum RSSI value to reject interference signals. In the five training sets for two APs and the eight training sets for three APs, the ED and PD thresholds were both -62dBm and -82dBm, respectively. Since these thresholds did not change, the impact of the PD and ED values ​​on the APs was not considered. However, their actual impact on the APs cannot be ignored. Due to the changes in the NAV12 thresholds, which were -82dBm and -62dBm, respectively, only the impact of the NAV threshold change was considered for threshold information.

[0131] We also introduce 0-1 variables to model the changes in the NAV threshold:

[0132]

[0133] Step S105: combining the extracted location identification information, the encoded discrete feature value, the calculated received signal strength indicator mean feature, and the encoded binary feature to form an input feature vector;

[0134] The topological location identification information extracted in step S101, the transport layer protocol discrete coding features generated in step S102 (a sequence of UDP=0, TCP=1), the RSSI mean features of all communication link pairs calculated in step S103, and the NAV threshold binary features (0 or 1) encoded in step S104 are used to generate the final input feature vector.

[0135] Step S106: setting an initial hyperparameter search range for the gradient boosting decision tree regression model;

[0136] The hyperparameters of the gradient boosting decision tree model include: subsampling rate of 0.8, L1 regularization weight of 0.01, L2 regularization weight of 0.1, number of decision trees of 100, minimum leaf node weight of 3, maximum tree depth of 5, learning rate of 0.05, and feature sampling ratio of 0.7.

[0137] Step S107: Use a grid search algorithm to traverse all combinations within the set initial hyperparameter search range, calculate the root mean square error corresponding to each combination through five-fold cross validation, and select the hyperparameter combination with the smallest root mean square error as the optimal hyperparameter combination;

[0138] In order to improve the accuracy of model prediction, the model parameters are adjusted and optimized during the training process, using the mean square error (MSE), root mean square error (RMSE) and R 2 Indicators, evaluation parameters and optimization results.

[0139] The model's predictive performance was comprehensively evaluated using multiple metrics, including mean squared error (MSE) and root mean square error (RMSE). MSE and RMSE reflect the average error between the model's predicted values ​​and the actual values, while the coefficient of determination measures the model's ability to explain data variation. These evaluation metrics comprehensively assess the model's fit and predictive accuracy, ensuring consistent and robust performance across both the training and validation sets, further verifying the model's effectiveness and reliability.

[0140] Five-fold cross-validation is a commonly used machine learning model evaluation technique used to estimate model performance and generalization ability. The original dataset is randomly and evenly divided into five equal-sized subsets. One of these subsets is selected as the test set, and the remaining four subsets are used as training sets. This process is repeated multiple times, with a different subset selected as the test set each time, ensuring that each subset has the opportunity to be used as the test set at least once.

[0141] Step S108: configuring a gradient boosting decision tree regression model using the obtained optimal hyperparameter combination, and performing model training using a training data set containing input feature vectors and corresponding access point effective transmission time series data;

[0142] AP sending time is affected by the network topology K n , service traffic P, RSSI between nodes (expressed as R) and threshold D have a common influence, and their mutual relationship can be expressed by the following mathematical model:

[0143] Seq time =F(K n ,P,R,D)(Formula 6)

[0144] Among them, the network topology K n It is determined by the relative location information loci of the network topology, the service flow P is determined by the protocol for sending information between AP nodes, the RSSI between nodes is determined by the data in Table 1 and Table 2, and the threshold D comes from the influence of the change of the NAV threshold.

[0145] After multiple rounds of training and parameter tuning, the root mean square error changes as the number of training rounds increases. Figure 7 As shown, the horizontal axis represents the number of gradient training, and the vertical axis represents the change of RMSE; Figure 7 As can be observed in Figure 2, the root mean square error gradually decreases with the increase in training rounds, indicating that the model's predictive performance is continuously improving. This trend indicates that with more training, the model is able to better fit the training data, thereby improving its performance on the validation set.

[0146] Step S109: input the input feature vector constructed by the scenario to be predicted into the trained gradient boosting decision tree regression model, and output access point effective transmission time series data.

[0147] As can be seen from formulas (3) and (4), throughput prediction relies on the effective transmission time series data (Seq_time) of the access point. Seq_time cannot be directly obtained from the test set, nor can it be replaced by other features. Therefore, this parameter must be provided by the gradient boosting decision tree regression model (XGBoost) trained by the above scheme.

[0148] In throughput prediction, it's crucial to pre-determine the modulation and coding scheme (MCS) and number of spatial streams (NSS). This is because they directly determine the theoretical physical layer transmission rate (PHYRate), a core input parameter of the throughput physical model. Without the MCS / NSS, relying solely on traditional features like RSSI to predict the PHYRate can lead to cumulative errors. The following describes a detailed prediction method for combining modulation and coding scheme data with the number of spatial streams. This method includes:

[0149] Step S201: obtaining received signal strength indication data and preset threshold information between at least two access points and a station;

[0150] This step acquires physical layer data in a high-density WLAN environment through actual measurement. Specifically, the data collected is the received signal strength indicator data and preset threshold information between at least two access points (APs) and their associated stations (STAs) within the deployment area. Received signal strength indicator (RSSI) data includes the measured signal strength values ​​(including sum, max, and mean statistics) recorded when the AP and STA communicate. Preset threshold information includes dynamic configuration parameters such as the packet detection threshold (PD), energy detection threshold (ED), and network allocation vector threshold (NAV).

[0151] Step S202: determining the synchronous / asynchronous communication state between the access points based on the dynamic relationship between the received signal strength indication data and the preset threshold information;

[0152] Each AP only sends downlink data to the associated specific STA (such as AP0→STA0, AP1→STA1). Figure 8 As shown in the figure, in the 2AP scenario, four directional communication links (including uplink and downlink) are formed, corresponding to four sets of RSSI data:

[0153] Downlink data sent by AP0 to STA0;

[0154] Uplink data sent by STA0 to AP0;

[0155] Downlink data sent by AP1 to STA1;

[0156] Uplink data sent by STA1 to AP1.

[0157] like Figure 9 As shown in FIG, in a wireless communication system including two access points (APs), the core basis for determining their communication status (synchronous or asynchronous) is to analyze the downlink data sent by each AP to its associated station (STA) (e.g. )and( )'s received signal strength indicator (RSSI) measured value.

[0158] The signal strength measured in the experiment includes three key indicators: sum (total), max (maximum) and mean (average). When judging the communication status between two access points (AP0 and AP1), the downlink data is used to determine the signal strength between the two access points. )and The RSSI value meets the threshold condition to determine:

[0159] like and If the measured signal strengths do not meet the conditions (i.e., the max value is not in the [PD, ED] interval and the mean value is not in the [NAV, ED] interval), it is determined to be in synchronous transmission state;

[0160] If there is only one downlink signal or If the conditions are met (i.e., the max value is located at [PD, ED] or the mean value is located at [NAV, ED]), it is determined to be in asynchronous transmission state;

[0161] If both downlink signals meet the conditions (i.e., their max values ​​are both located at [PD, ED] or their mean values ​​are both located at [NAV, ED]), then it is determined to be a synchronous-asynchronous mixed state.

[0162] By dynamically comparing the peak strength (max) and overall level (mean) of the downlink signal with the preset thresholds (PD, ED, NAV), the communication status classification between the two APs is finally output.

[0163] In the 3AP scenario ( Figure 10 ), expanded to 6 directed links (bidirectional interaction between AP0-STA0, AP1-STA1, and AP2-STA2), with Figure 10 For the case where there are 3 APs, we select 2 APs from the 3 APs and determine the synchronous / asynchronous mixed state according to the analysis method for 2 APs. Figure 11 As shown, for a communication scenario including three access points, the method for determining the synchronous / asynchronous communication state between the access points includes:

[0164] First, three access points are marked as the first access point, the second access point, and the third access point;

[0165] Sequentially select two access points to form access point pairs, including: a first access point and a second access point combination, a first access point and a third access point combination, and a second access point and a third access point combination; and perform the following determination operations on each access point pair:

[0166] Synchronous transmission status determination: When the API downlink data signal Downlink data signal with APJ When the received signal strength indicator value does not meet any of the following conditions:

[0167] 1. The maximum signal strength is within the interval formed by the packet detection threshold and the energy detection threshold;

[0168] 2. If the average signal strength value is within the interval formed by the network allocation vector threshold and the energy detection threshold, it is determined that API and APJ are in a synchronous transmission state.

[0169] Asynchronous transmission status determination: When only one of the downlink data signals of API and APJ meets any of the following conditions:

[0170] 1. The maximum signal strength reaches the packet detection threshold requirement;

[0171] 2. The average signal strength reaches the network allocation vector threshold requirement;

[0172] 3. It is determined that the API and APJ are in asynchronous transmission state.

[0173] Mixed transmission status determination: When the received signal strength indicator values ​​of the downlink data signal of API and the downlink data signal of APJ both meet any of the following conditions:

[0174] 1. The maximum signal strength reaches the packet detection threshold requirement;

[0175] 2. The average signal strength reaches the network allocation vector threshold requirement;

[0176] 3. It is determined that API and APJ are in a synchronous and asynchronous mixed transmission state.

[0177] Finally, the communication status categories of all access point pairs are combined to generate the overall judgment result of the network communication status including three access points, such as Figure 12 shown.

[0178] Step S203: Based on the synchronous / asynchronous communication state, classify the signal sent by the adjacent access point to the station as an interference signal, and classify the signal sent by the adjacent station as environmental noise.

[0179] For the case where there are 2 APs, when classifying, we first analyze the characteristics of the three states: synchronous transmission, asynchronous transmission, and synchronous / asynchronous mixed transmission. In synchronous transmission, APs listen to each other in real time, and most of the interference when STA receives data comes from the environmental noise floor. In asynchronous transmission, APs do not listen to each other at all, and STA ignores the environmental noise floor and only receives interference signals. In synchronous / asynchronous mixed transmission, APs listen to each other in real time and do not listen to each other alternately, and STA receives environmental noise floor and interference signals. Based on data analysis and knowledge of communication principles, for a certain STA node, the signal from the adjacent AP node is the interference signal, and the signal from other adjacent STA sites is the environmental noise floor, such as Figure 13 shown.

[0180] For the case where there are 3 APs, the classification of interference signals and ambient noise is based on the same method as for the case where there are 2 APs, thus obtaining the classification of interference signals and ambient noise. Figure 14 shown.

[0181] Step S204: Calculate the signal-to-noise ratio of the site based on the signal classification result;

[0182] For the scenario where there are two APs, based on the classification of interference signals and ambient noise, the STA signal-to-noise ratio (SINR) calculation model is dynamically adjusted according to the synchronous / asynchronous communication status between the access points. The SINR calculation method is:

[0183] SINR=P signal -(P noise +P inter )(Formula 7)

[0184] Among them, P signal Represents the useful signal (corresponding to Figure 13 The black solid line in the figure), P noise Represents the ambient noise (corresponding to Figure 13 Green dashed line in the figure), P inter Indicates the interference signal (corresponding to Figure 13 According to the three cases of synchronous / asynchronous states between two APs, the calculation formula of signal-to-noise ratio is modified as follows: Figure 15 .

[0185] Synchronous communication state: When two access points (AP0 and AP1) are in synchronous transmission, the signal-to-noise ratio calculation of the station (STA) only considers the ambient noise power (P noise ), the formula is: SINR = P Signal -Pnoise .

[0186] Asynchronous communication state: When the two access points are in asynchronous transmission, the signal-to-noise ratio calculation only considers the interference signal power (P inter ), the formula is: SINR = P Signal -P inter .

[0187] Mixed communication state: When two access points are in synchronous and asynchronous mixed transmission, it is necessary to aggregate the linear power value (in mW) of the ambient noise floor and the interference signal at the same time. The formula is: SINR = P Signal -P noise -P inter .

[0188] The network topology with three APs is essentially a directed complete graph consisting of six nodes, with a total of 30 directed edges. Therefore, the difference in SINR calculation compared to the one with two APs lies in the aggregation process of multiple signals.

[0189] Taking the example of 3 APs interacting synchronously and asynchronously, when summing the signals in the following formula, it is necessary to first sum each sub-signal (P in dBm) inter or P noise ) is converted into linear power in mW, then the linear power of all interference signals is added together and finally converted back to dBm. Therefore, the formula for calculating the cumulative strength of n signals is:

[0190]

[0191] Among them, P total_n Indicates the cumulative value of the intensity of n signals, P1....P n Indicates the signal strength of each of the n signals. Figure 14 Taking the station STA0 in the example, the signal-to-noise ratio calculation formula of the station at a certain moment is:

[0192]

[0193] in, represents the effective signal strength sent from AP0 to STA0 (black solid line), P env1 represents the ambient noise from STA1 (red dashed line), P inter1 represents the interference signal from AP1 (green dashed line), P inter2 Represents the interference signal from AP2 (green dashed line).

[0194] Step S205: inputting the signal-to-noise ratio, synchronous / asynchronous communication status, and access point transmit power into a gradient boosting decision tree model;

[0195] Based on the prediction model with 2 APs and 3 APs, the calculated STA signal-to-noise ratio value is combined with some basic test information as input to obtain the output (MSC, NSS). Figure 16 The gradient boosting decision tree model training process shown in the figure, where the model input layer is composed of three types of features: site signal-to-noise ratio features (env_SINR: signal-to-noise ratio value calculated based on environmental noise floor power, ferr_SINR: signal-to-noise ratio value calculated based on interference signal power), access point status features (categ_0, categ_1, categ_2: three-dimensional classification variables characterizing the synchronous / asynchronous communication status between APs), and test environment basic features (RSSI: measured value of received signal strength indicator, eirp: access point transmit power, nav: network allocation vector threshold value, loc_id: test location identifier, ap_id: access point device identifier).

[0196] Step S206: Output the modulation and coding scheme and spatial stream number prediction result of the target access point through the gradient boosting decision tree model.

[0197] The model output layer is the target prediction item:

[0198] Modulation and Coding Scheme (MCS): predicts the most commonly used modulation and coding scheme after AP adaptive adjustment stabilizes;

[0199] Number of Spatial Streams (NSS): This value indicates the number of spatial streams most frequently used by the AP after adaptive adjustment stabilizes.

[0200] according to Figure 16 The gradient boosting decision tree algorithm model is trained by the process to predict the (MSC, NSS) combination of two environments with two APs and three APs.

[0201] For the five training datasets with an AP number of 2, 5-fold cross validation was used to fully train the model and also verify the effectiveness of the model.

[0202] exist Figure 17 In the graph, the horizontal axis represents the actual NSS or MOS value, and the vertical axis represents the model's predicted value. When a node is located on a diagonal line passing through the origin, the prediction is accurate. The color of the node indicates the number of training data points at that location. Darker points on the diagonal line indicate a high prediction accuracy. Therefore, it is intuitively clear that the model is highly accurate.

[0203] Secondly, the prediction accuracy of the statistical model is 86.15% ( Figure 18(a)), MSE is 0.6518; 376 NSS are accurately predicted, with an accuracy of 96.41% ( Figure 18 (b)), the MSE is 0.5128. In addition, the trained gradient boosting decision tree algorithm model can give the importance ranking of influence (MOSNSS).

[0204] Finally, through Figure 19 According to the importance ranking of (a) and (b), we can draw the following conclusions:

[0205] The STA's signal-to-noise ratio has the greatest impact on (MOS, NSS). Reason: The environmental noise floor SINR (env_SINR) and the interference signal SINR (ferr SINR) ranked first and third, respectively, indicating that the STA's signal-to-noise ratio has the greatest impact on (MCS, NSS).

[0206] In the signal-to-noise ratio (SNR) component, noise from other STAs is more important than noise from other APs. Reason: When ranking the factors influencing MCS, the environmental noise floor increases the impact of interference noise by approximately 50%. When ranking the factors influencing NSS, the environmental noise floor increases the impact of interference noise by approximately 160%. The AP's transmit power has a more significant impact on (MCS, NSS). Reason: When ranking the factors influencing both MCS and NSS, AP transmit power (eirp) ranks second. Threshold information and the synchronization / asynchrony status information determined by the threshold have a smaller impact on (MCS, NSS). Reason: Changes in the NAV threshold and synchronization / asynchrony status information (represented by category_0, category_1, and category_2) are ranked low in terms of influence.

[0207] By implementing a 5-fold cross-validation strategy on the loc_nav.csv dataset (8 training_set_APs with 3 APs and 3 APs), we generated 858 training data points. This approach not only ensures that the model is fully trained and covers data diversity, but also rigorously evaluates the model's performance on different subsets through cross-validation, thereby verifying the model's effectiveness and generalization ability.

[0208] from Figure 20 The displayed data point distribution cleverly reflects their density and importance, while their positions directly correspond to the relationship between the model's predicted (MCS, NSS) values ​​and the actual measured values. Most data points are closely clustered around the actual measured values ​​and are evenly distributed, indicating good consistency between the model's predictions and the measured data. Therefore, we can conclude that the model demonstrates high accuracy in predicting (MOS, NSS).

[0209] Secondly, the prediction accuracy of the statistical model is that 830 out of 858 training data accurately predict NSS, with an accuracy of 96.74% ( Figure 21 (b)), MSE is 0.04315; 559 data accurately predict MCS with an accuracy of 65.16% ( Figure 21 (a)), the MSE is 1.8949. Although the prediction accuracy is not high when the number of APs is 3, the MSE is low, which shows that although there is a small deviation in the prediction, the prediction result is still relatively accurate. Figure 21 In addition, the trained gradient boosting decision tree algorithm model can give the importance ranking of the impact (MCS, NSS) of the number of APs in three environments, as shown in Figure 22 As shown in (a) and (b).

[0210] According to the importance ranking of the three impacts (MCS, NSS) based on the number of APs, the following conclusions can be drawn:

[0211] The STA's signal-to-noise ratio has the greatest impact on (MOS, NSS). Reason: The environmental noise floor SINR (env_SINR) and the interference signal SINR (ferr SINR) ranked first and second, respectively, indicating that the STA's signal-to-noise ratio has the greatest impact on (MCS, NSS).

[0212] Interference noise is more important in an environment with three APs than in an environment with two APs. Reason: In the ranking of factors affecting MCS and NSS, the importance of ambient noise and interference noise is roughly the same, with ambient noise being slightly more important than interference noise.

[0213] AP transmit power has a significant impact on MCS and NSS. Reason: AP transmit power (eirp) ranks fourth and third among the factors influencing MOS and NSS, respectively. Threshold information and the synchronization / asynchrony status information determined by the threshold have a smaller impact on MCS and NSS. Reason: Changes in the NAV threshold and synchronization / asynchrony status information (represented by category_0, category_1, and category_2) are ranked low in terms of influence.

[0214] Using the gradient boosting decision tree algorithm for training and 5-fold cross-validation, the results show that when the number of APs is 2, the NSS prediction accuracy is 96.4% and the MOS prediction accuracy is 86.2%. When the number of APs is 3, the NSS prediction accuracy is 96.7% and the MOS prediction accuracy is 65.2%. The model test shows that the accuracy of (MOS, NSS) prediction is higher when the number of APs is 2 than when the number of APs is 3. Although the prediction accuracy is not high when the number of APs is 3, the mean squared error (MSE) of the prediction is only 1.89, which is relatively small. The majority of the incorrectly predicted MCSs are not much different from the true MCSs, indicating that the prediction model is effective.

[0215] It can be understood that the method of the present invention proposes a hierarchical modeling framework for the throughput prediction problem in a high-density WLAN environment: first, an AP transmission time prediction model is established based on XGBoost, and the network topology, service traffic (protocol type), threshold (NAV) and RSSI characteristics between nodes are input. Through grid search parameter adjustment and five-fold cross-validation, the conclusion is drawn that the service traffic has the strongest impact; then, the SINR feature is constructed (integrating synchronous / asynchronous state discrimination, interference signal classification and environmental noise floor modeling), and combined with XGBoost to predict the MCS / NSS combination. The MCS / NSS accuracy reaches 86.2% / 96.4% in the 2AP scenario, and the NSS remains at 96.7% under 3AP, but the MCS drops to 65.2% (attributed to the fact that RSSI only reflects large-scale channel changes), and SINR is verified to be the most critical feature; finally, the first two outputs are integrated, and XGBoost is first tried to predict the throughput, and then a physical model correction strategy is proposed: the basic value is calculated based on the theoretical formula, and then the correction coefficient is introduced through linear regression to significantly reduce the MSE.

[0216] According to another aspect of an embodiment of the present application, an electronic device is provided, including a processor and a memory, wherein the processor is configured to implement the steps of the method when executing a computer program stored in the memory.

[0217] In the above embodiments of the present invention, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0218] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only exemplary. For example, the division of the units can be a logical function division. In actual implementation, there may be other division methods, such as 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 interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0219] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, 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.

[0220] If the integrated unit is implemented in the form of 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 the present invention, 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. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, etc. Various media that can store program codes.

[0221] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A wireless local area network throughput prediction method, characterized in that: The steps include: Obtain network topology data, signal strength indicator data, channel access mechanism data, modulation and coding scheme and spatial stream combination data, packet error rate data, total test duration data, and predicted access point effective transmission time series data; Substitute the acquired modulation and coding scheme combination data, spatial stream number data, packet error rate data, total test duration data, and access point effective transmission time series data into the initial physical model to calculate the initial throughput prediction value; Using the linear regression algorithm, the calculated initial throughput prediction value is fitted with the measured throughput to obtain the slope correction coefficient and intercept correction term; Applying the obtained slope correction coefficient and intercept correction term to the initial physical model to construct a corrected physical model; Use cross-validation method to train and verify the constructed modified physical model; Based on the modified physical model after training and verification, the obtained network topology data, signal strength indicator data, and channel access mechanism data are input to predict the overall network throughput.

2. The wireless local area network throughput prediction method according to claim 1, wherein: The method for predicting effective transmission time series data of an access point includes: Obtaining the network topology data of the target wireless local area network and extracting the location identification information of the access points and stations therefrom; Collecting service flow data of the wireless local area network and encoding the transport layer protocol type into discrete feature values; Measuring received signal strength indicator values ​​between all nodes in the network topology data and calculating received signal strength indicator mean characteristics of all communication link pairs; Reading a network allocation vector threshold value configured in the wireless local area network and encoding it into a binary feature; Combining the extracted location identification information, the encoded discrete feature value, the calculated received signal strength indication mean feature, and the encoded binary feature to form an input feature vector; Set the initial hyperparameter search range for the gradient boosted decision tree regression model; A grid search algorithm is used to traverse all combinations within the set initial hyperparameter search range, and the root mean square error corresponding to each combination is calculated through five-fold cross validation. The hyperparameter combination with the smallest root mean square error is selected as the optimal hyperparameter combination. The optimal hyperparameter combination is used to configure a gradient boosting decision tree regression model, and the model is trained using a training dataset containing input feature vectors and corresponding access point effective transmission time series data; The input feature vector constructed by the scenario to be predicted is input into the trained gradient boosting decision tree regression model, and the access point effective transmission time series data is output.

3. The wireless local area network throughput prediction method according to claim 2, wherein: Methods for encoding transport layer protocol types into discrete feature values ​​include: When the protocol type is User Datagram Protocol, the value is 0; When the protocol type is Transmission Control Protocol, the value is 1.

4. The wireless local area network throughput prediction method according to claim 2, wherein: The five-fold cross validation is performed as follows: The training dataset is randomly divided into five mutually exclusive subsets; Select one subset as the validation set and the remaining four subsets as the training set; Train the model under each hyperparameter combination and calculate the root mean square error on the validation set; After five cycles, the average root mean square error is taken as the evaluation index.

5. The wireless local area network throughput prediction method according to claim 1, wherein: The method for predicting the combined data of the modulation and coding scheme and the number of spatial streams includes: Obtaining received signal strength indication data and preset threshold information between at least two access points and a station; Determining the synchronous / asynchronous communication state between the access points according to the dynamic relationship between the received signal strength indication data and the preset threshold information; Based on the synchronous / asynchronous communication state, classify the signal sent by the adjacent access point to the station as an interference signal, and classify the signal sent by the adjacent station as environmental noise; Calculate the signal-to-noise ratio of the site based on the signal classification results; Inputting the signal-to-noise ratio, synchronous / asynchronous communication state and access point transmit power into a gradient boosting decision tree model; The gradient boosting decision tree model is used to output the prediction results of the modulation and coding scheme of the target access point and the spatial stream number combination data.

6. The wireless local area network throughput prediction method according to claim 5, wherein: Methods for determining the synchronous / asynchronous communication status between access points include: For the downlink data signal of each access point, extract the maximum value and average value of its received signal strength indicator data; When the maximum values ​​of the received signal strength indicator data of the downlink data signal of the first access point and the downlink data signal of the second access point do not reach the packet detection threshold and the average values ​​of the received signal strength indicator data do not reach the network allocation vector threshold, it is determined that the two access points are in a synchronous communication state; When the maximum value of the received signal strength indicator data of only one of the downlink data signal of the first access point and the downlink data signal of the second access point reaches the packet detection threshold or the average value of the received signal strength indicator reaches the network allocation vector threshold, it is determined that the two access points are in an asynchronous communication state; When the maximum values ​​of the received signal strength indication data of the downlink data signal of the first access point and the downlink data signal of the second access point both reach the packet detection threshold or the average values ​​of the received signal strength indication data both reach the network allocation vector threshold, it is determined that the two access points are in a mixed communication state.

7. The wireless local area network throughput prediction method according to claim 5, wherein: The signal classification rules are: Interference signals are defined as signals from non-associated access points; The ambient noise floor is defined as the signal from non-target sites.

8. The wireless local area network throughput prediction method according to claim 5, wherein: Calculate the signal-to-noise ratio value of the site: When APs are synchronized, the signal-to-noise ratio calculation ignores the ambient noise floor intensity. When APs are in an asynchronous state, the signal-to-noise ratio calculation ignores the interference signal strength; When APs are in a mixed synchronous and asynchronous state, this value aggregates the linear power of the ambient noise floor and the interference signal.

9. The wireless local area network throughput prediction method according to claim 8, wherein: Methods for aggregating the linear power values ​​of environmental noise floor and interference signals include: Convert all signal strength values ​​to be aggregated from dBm to mW. Sum the converted mW unit values; Convert the summation result back to dBm units.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the wireless local area network throughput prediction method according to any one of claims 1 to 9 are executed.

Images