Power management method for wireless local area network module of computer

Abstract

The invention relates to the technical field of wireless local area network power management, in particular to a power management method of a computer wireless local area network module. The method comprises the following steps of: acquiring network traffic data and application program state data by utilizing network equipment management software, constructing a traffic prediction model by utilizing an autoregression moving average model, defining a network traffic fuzzy set and an application program demand fuzzy set, establishing a fuzzy rule, determining an input membership degree, and determining an application program demand fuzzy set; and calculating an output membership degree according to the rule trigger intensity, converting the fuzzy mode membership degree into a specific voltage parameter by using a gravity center method, and regulating and controlling the power supply. According to the method, the flow prediction model is constructed, the working mode of the wireless local area network module is dynamically adjusted by using the clustering analysis method, and the fuzzy mode membership degree is converted into the specific voltage parameter by using the gravity center method, so that the power management of the wireless local area network module is finely controlled, and the power consumption is reduced.

Description

Technical Field

[0001] The invention relates to the technical field of wireless local area network power management, in particular to a power management method for a computer wireless local area network module. Background Art

[0002] The computer's wireless LAN module is a hardware component that enables the computer to connect to a local area network wirelessly. Its main function is to receive and send wireless signals, so that the computer can communicate with a wireless router or other wireless access point. The wireless LAN module is a relatively power-consuming component. When it is in operation, especially when performing high-intensity data transmission or continuously searching for surrounding wireless networks, it consumes a lot of power. Through effective power management, the power consumption of the wireless LAN module can be reduced when the wireless network is not needed, which can not only extend the battery life of the device, but also optimize system resources and performance.

[0003] At present, when the power management of the wireless LAN module of the computer is performed, it is simply controlled by a power gating circuit. When the wireless LAN module is in an idle state, the power gating circuit can cut off the power supply of some circuits of the module. When there is a new data transmission demand, the power supply is quickly restored and the corresponding circuit is started. The working mode of the wireless LAN module cannot be adjusted according to the size of the data transmission, resulting in power consumption. In order to be able to manage the power of the wireless LAN module in real time according to the network traffic data and the application status data, and specifically adjust the relevant parameters of the power supply, improve the energy utilization rate, and at the same time, ensure the normal working needs of the computer, therefore, we propose a power management method for the wireless LAN module of the computer. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that when the wireless LAN module of a computer performs power management, the working mode of the wireless LAN module cannot be adjusted according to the size of data transmission. In order to be able to manage the power of the wireless LAN module in real time according to network traffic data and application status data, and specifically adjust the relevant parameters of the power supply, dynamically adjust the working mode of the wireless LAN, and improve energy utilization.

[0005] To achieve the above object, the present invention provides a power management method for a computer wireless local area network module, comprising the following steps:

[0006] S1. Collect network traffic data and application status data using network device management software, and build a traffic prediction model using an autoregressive moving average model;

[0007] S2, define network traffic fuzzy set and application requirement fuzzy set, and establish fuzzy rules;

[0008] S3, determining the input membership according to the collected network traffic data and application demand data, and calculating the output membership according to the rule triggering strength;

[0009] S4. Use the centroid method to convert the fuzzy mode membership into specific voltage parameters and regulate the power supply.

[0010] As a further improvement of the technical solution, the S1 uses an autoregressive moving average model to build a traffic prediction model, and the method steps are as follows:

[0011] S1.1.1, Data cleaning and processing: Clean the historical network traffic data collected from the wireless LAN module and use the differential method to process the data smoothly;

[0012] S1.1.2. Determine model parameters: determine the autoregressive order p of the model using the autocorrelation function, determine the moving average order q using the Yule-Walker equation, and determine the differential order d using the autocorrelation function graph and the partial autocorrelation function graph;

[0013] S1.1.3, Model fitting: Use the least squares method to estimate the model parameters and test whether the residuals of the model meet the white noise assumption;

[0014] S1.1.4, Traffic prediction stage: Use the fitted autoregressive moving average model to predict traffic, and the formula is:

[0015]

[0016] in, is the autoregressive operator, θ(B) is the moving regression operator, B is the lag operator, Y t is time series data, ε t is white noise, and d is the difference order.

[0017] As a further improvement of the technical solution, S1.1.2 uses the autocorrelation function to determine the autoregressive order p of the model, and the formula is:

[0018]

[0019] Among them, p k is the autoregressive order of the lag k period, T is the length of the time series, y t is a time series, is the time series y t The mean of , k is the number of lags, y t-k is the time series from k to t.

[0020] As a further improvement of the technical solution, when fitting the S1.1.3 model, the normality of the residuals is tested using Shapiro-Wilk, and if the residuals do not conform to the normal distribution, the model parameters are adjusted.

[0021] As a further improvement of the technical solution, the S2 establishes fuzzy rules, and the method steps are as follows:

[0022] S2.1.1, Cluster analysis process: Use K-Means clustering algorithm to cluster network traffic data and application demand data;

[0023] S2.1.2, define the scope of fuzzy sets: analyze each cluster to determine the network state it represents, and determine the scope of fuzzy sets based on the data distribution of cluster centers and members;

[0024] S2.1.3. Establish fuzzy rules: Establish fuzzy rules based on the network status represented by the cluster and the actual power management requirements, add device status factors, and refine the rules.

[0025] As a further improvement of the present technical solution, in the cluster analysis process S2.1.1, the elbow rule is used to determine the appropriate number of clusters, measure the relationship between the compactness of clusters and the number of clusters, and find the turning point where the cluster change trend changes from a sharp decline to a gentle decline.

[0026] As a further improvement of the technical solution, the step S3 determines the input membership, and the steps are as follows:

[0027] S3.1.1. Determine the parameters of the Gaussian membership function: the statistical mean of the variable corresponding to the fuzzy set is used as the center position of the Gaussian membership function, and the standard deviation of the fuzzy set data is used to determine the width of the Gaussian membership function;

[0028] S3.1.2, Calculate the triggering strength of the rule: determine the membership of the network traffic and application requirements in the corresponding fuzzy set, and select the minimum value of the input membership as the triggering strength of the rule;

[0029] S3.1.3. Calculate output membership: When the rule is triggered, determine the membership of the output fuzzy set based on the conclusion of the rule.

[0030] As a further improvement of the present technical solution, when determining the Gaussian membership function parameters in S3.1.1, the standard deviation is adjusted by the expected membership change rate to adjust the shape of the membership function.

[0031] As a further improvement of the present technical solution, when calculating the output membership in S3.1.3, if there are multiple rules affecting the membership of the output fuzzy set, the maximum value method is used to determine the final output membership.

[0032] As a further improvement of the technical solution, the S4 membership is converted into a specific voltage parameter, wherein the contribution of each fuzzy set to the power supply parameter is calculated, and the final voltage is obtained according to the voltage conversion formula, wherein the voltage conversion formula is as follows:

[0033]

[0034] Where V is the voltage, V min is the minimum value of the lower limit of the voltage range in all modes, ∑ i μ i is the sum of all memberships, C i is the voltage contribution, and n is the number of memberships.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. The power management method of the computer wireless LAN module collects network traffic data and application status data through network device management software, and uses the autoregressive moving average model to build a traffic prediction model to predict the traffic usage of the computer. According to the predicted network traffic and application requirements, the cluster analysis method is used to dynamically adjust the range of the corresponding fuzzy set and establish corresponding fuzzy rules, and then dynamically adjust the working mode of the wireless LAN module according to the transmission of traffic data;

[0037] 2. Determine the input membership based on the collected network traffic data and application demand data, and calculate the output membership based on the rule trigger strength. Then use the center of gravity method to convert the fuzzy pattern membership into specific voltage parameters, and regulate the power supply. Through power distribution, adjust the voltage of the wireless LAN module to a specific value, realize refined control of the power management of the wireless LAN module, and reduce energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0039] Figure 2 It is a schematic flow chart of the step S1 of the present invention;

[0040] Figure 3 This is a schematic flow chart of the steps of S2 of the present invention;

[0041] Figure 4 It is a schematic flow chart of the steps of S3 of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] At present, when the wireless LAN module of a computer performs power management, the working mode of the wireless LAN module cannot be adjusted according to the size of data transmission. In order to be able to manage the power of the wireless LAN module in real time according to network traffic data and application status data, and specifically adjust the relevant parameters of the power supply, dynamically adjust the working mode of the wireless LAN, and improve energy utilization.

[0044] Therefore, the present invention proposes to collect network traffic data and application status data through network equipment management software, and use the autoregressive moving average model to build a traffic prediction model, define network traffic fuzzy sets and application demand fuzzy sets, and establish fuzzy rules, determine the input membership based on the collected network traffic data and application demand data, and calculate the output membership based on the rule trigger strength, use the center of gravity method to convert the fuzzy pattern membership into a specific voltage parameter, and regulate the power supply to achieve refined control of the power management of the wireless LAN module and reduce power consumption.

[0045] The details are as follows:

[0046] See also Figure 1 As shown, the present invention provides a power management method for a computer wireless local area network module, comprising the following steps:

[0047] S1. Collect network traffic data and application status data using network device management software, and build a traffic prediction model using an autoregressive moving average model;

[0048] S2, define network traffic fuzzy set and application requirement fuzzy set, and establish fuzzy rules;

[0049] S3, determining the input membership according to the collected network traffic data and application demand data, and calculating the output membership according to the rule triggering strength;

[0050] S4. Use the centroid method to convert the fuzzy mode membership into specific voltage parameters and regulate the power supply.

[0051] like Figure 2 As shown, S1 uses an autoregressive moving average model to build a traffic prediction model, and the steps are as follows:

[0052] S1.1.1, Data cleaning and processing: Clean the historical network traffic data collected from the wireless LAN module and use the differential method to process the data smoothly;

[0053] S1.1.2. Determine model parameters: determine the autoregressive order p of the model using the autocorrelation function, determine the moving average order q using the Yule-Walker equation, and determine the differential order d using the autocorrelation function graph and the partial autocorrelation function graph;

[0054] S1.1.3, Model fitting: Use the least squares method to estimate the model parameters and test whether the residuals of the model meet the white noise assumption;

[0055] S1.1.4, Traffic prediction stage: Use the fitted autoregressive moving average model to predict traffic, and the formula is:

[0056]

[0057] in, is the autoregressive operator, θ(B) is the moving regression operator, B is the lag operator, and Y t is time series data, ε t is white noise, and d is the difference order.

[0058] Collect historical network traffic data of the wireless LAN module. This data should be arranged in time series. For example, the size of network traffic (which can be data transmission volume, number of data packets, etc.) should be recorded at fixed time intervals (such as every minute, every 5 minutes, etc.). The time span of the collected data should be long enough to include different network usage scenarios and traffic patterns, such as collecting data for at least the past week or month.

[0059] Clean the data to remove outliers. Outliers may be caused by network failures, temporary high-load applications, or monitoring errors. Outliers can be identified through statistical methods. For example, if the difference between a data point and adjacent data points exceeds a certain multiple of the standard deviation, the data point is considered an outlier and is appropriately processed (such as replacing it with the average of adjacent data points).

[0060] Process the data for stability. Network traffic data may have non-stationary characteristics such as seasonality and trend. The commonly used method is the difference method. For example, perform the first-order difference on the original data (i.e., subtract the previous data point from the next data point) to observe whether the data after the difference is stable. If it is still not stable, you can continue to perform the second-order difference or other appropriate transformation until the data is stable. Stable data is more easily processed by the autoregressive moving average model.

[0061] In order to determine the autoregressive order p more accurately, S1.1.2 uses the autocorrelation function to determine the autoregressive order p of the model, and the formula is:

[0062]

[0063] Among them, p k is the autoregressive order of the lag k period, T is the length of the time series, y t is a time series, is the time series y t The mean of , k is the number of lags, y t-k is the time series from k to t.

[0064] Suppose we have a simple time series, y = [y1, y2, y3, y4, y5]. First, calculate its mean:

[0065]

[0066] Calculate the autocorrelation coefficient p1 for lag k = 1, the numerator is When t = 2, Similarly, calculate the scores of each item from t = 2 to t = 5 and sum them up. The denominator is Calculate p1 according to the autoregressive order formula. Repeat the above calculation process for different lag values ​​k to obtain the complete autocorrelation function.

[0067] For an autoregressive model of order p, in, is the autoregressive coefficient, ε t is white noise. According to the Yule-Walker equation, the partial autocorrelation coefficient (u-order partial autocorrelation coefficient) satisfies the following equations:

[0068]

[0069] For u = 1, Through the order equation system, we get and By analogy, higher-order partial autocorrelation coefficients q can be calculated.

[0070] Once the values ​​of p and q are determined, the selected ARIMA(p,q) model can be used to fit the preprocessed network traffic data. During the fitting process, the least squares method is used to estimate the parameters of the model;

[0071] The goal of the least squares method is to find a set of parameters that minimizes the sum of squared errors between the model predictions and the actual observations.

[0072] For the linear regression model y = β0 + β1x1 + ... + β n x n +ε, where y is the dependent variable, x i is the independent variable, β i is the parameter to be estimated, and ε is the error term. The error sum of squares is Among them, m is the number of samples, y i is the actual observed value, is the model's predicted value.

[0073] Take the partial derivative of the sum of squared errors:

[0074] Solving the equation yields: Then we get the parameters of the model

[0075] In order to adjust the model parameters, when fitting the S1.1.3 model, the normality of the residuals was tested using the Shapiro-Wilk method. If the residuals did not conform to the normal distribution, the model parameters were adjusted.

[0076] The Shapiro-Wilk test is a statistical test method based on sample data to test whether it comes from a normal distribution population. It measures the degree of deviation of the data from the normal distribution by calculating a statistic based on the order and covariance matrix of the sample data. The null hypothesis is that the data follows a normal distribution, and the alternative hypothesis is that the data does not follow a normal distribution.

[0077] If the residual value is greater than the set significance level (usually 0.05), the null hypothesis cannot be rejected, that is, the residual can be considered to follow a normal distribution; when the value is less than or equal to the significance level, the null hypothesis is rejected, and the residual is considered to not follow a normal distribution. For example, if the residual value is 0.03, it means that at the significance level, the residual does not follow a normal distribution, and further data transformation (such as logarithmic transformation, Box-Cox transformation, etc.) or reconsidering the model setting may be required.

[0078] Use the fitted ARIMA model to predict traffic. Assuming that the ARIMA (p, q) model has been fitted and there is network traffic data for n past time points, to predict the traffic at m future time points, first input the past data into the model to get the current state of the model, and then gradually calculate the predicted values ​​at future time points according to the model equations, using the formula:

[0079]

[0080] in is the autoregressive operator, θ(B) is the moving regression operator, B is the lag operator, and Y t is time series data, εt is white noise, d is the difference order. The predicted value at the future time point is obtained by recursive calculation.

[0081] like Figure 3 As shown, S2 establishes fuzzy rules, and the method steps are as follows:

[0082] S2.1.1, Cluster analysis process: Use K-Means clustering algorithm to cluster network traffic data and application demand data;

[0083] S2.1.2, define the scope of fuzzy sets: analyze each cluster to determine the network state it represents, and determine the scope of fuzzy sets based on the data distribution of cluster centers and members;

[0084] S2.1.3, Establish fuzzy rules: Establish fuzzy rules based on the network status represented by the cluster and the actual power management requirements, add device status factors, and refine the rules;

[0085] Among them, the K-Means clustering algorithm divides the data set into K clusters, each cluster has a center (centroid), takes the normalized network traffic data and related variables as input, and runs the selected clustering algorithm. For example, the K-Means clustering algorithm is used to divide the data into 3 clusters. After the clustering is completed, each data point is assigned to a cluster, and the center and members of each cluster are obtained. Each cluster is analyzed to determine the network status it represents. For example, a data point in a cluster may have a high data transmission rate, a large number of data packets, and the corresponding device status is that the screen is on and running a high-demand application. This cluster can be defined as a "high traffic-high demand" fuzzy set;

[0086] Determine the range of the fuzzy set based on the data distribution of the cluster center and members. For the "high traffic-high demand" fuzzy set, take the data transmission rate as an example to calculate the minimum, maximum and average data transmission rate in the cluster. If the average value is 7MB / s, the minimum value is 5MB / s, and the maximum value is 10MB / s, this range can be used as the range of the traffic part in the "high traffic-high demand" fuzzy set. For other variables (such as application requirements), a similar method is used to determine the range;

[0087] Fuzzy rules are established based on the network status represented by the clusters and the actual power management requirements. For example, for the "high traffic-high demand" fuzzy set, a rule can be established: "If the network status is 'high traffic-high demand', then the wireless LAN module enters active mode."

[0088] In order to determine the number of clusters more accurately, S2.1.1 In the cluster analysis process, the elbow rule is used to determine the appropriate number of clusters, measure the relationship between the compactness of clusters and the number of clusters, and find the turning point where the cluster change trend changes from a sharp decline to a gentle decline;

[0089] The core idea of ​​the elbow rule is to measure the relationship between the closeness of clusters and the number of clusters. When the number of clusters increases, the data points in each cluster will be closer, and the sum of the distances from the data points to the center of the cluster to which they belong (usually measured by the sum of squared errors SSE) will gradually decrease. However, as the number of clusters continues to increase, the rate of decrease in SSE will gradually slow down. What we are looking for is the turning point where this trend changes from a sharp decline to a gentle decline. The number of clusters corresponding to this point is more appropriate, just like the position where a person's elbow bends;

[0090] Select a range of cluster numbers, usually starting with a smaller value, such as k=2. For each k value, execute the K-Means clustering algorithm. After each clustering is completed, calculate the sum of squared errors (SSE). Use the number of clusters k as the horizontal axis and the corresponding SSE value as the vertical axis to draw a curve. Observe the shape of the curve and find the bending point of the curve, which is the elbow. At the beginning of the curve, as k increases, the SSE will drop rapidly; after the elbow, the rate of SSE decrease will slow down significantly. The k value corresponding to this elbow is a more appropriate number of clusters;

[0091] Assume that we have a set of network traffic and device status data of wireless LAN modules, and perform cluster analysis on them to determine the appropriate number of power management mode clusters. When k=2, the calculated SSE is 1000; when k=3, the SSE drops to 600; when k=4, the SSE is 400; when k=5, the SSE is 300; when k=6, the SSE is 250; when k=7, the SSE is 220; when k=8, the SSE is 200;

[0092] After drawing the curve, we can find that when k=4, the curve is more obviously curved. From k=4 to k=5, the decrease in SSE is smaller than before. Therefore, according to the elbow rule, the appropriate number of clusters may be 4. This means that the data can be divided into 4 clusters to establish fuzzy sets and rules for power management. For example, these 4 clusters may correspond to four different network and device status combinations: high traffic and high demand, high traffic and low demand, low traffic and high demand, and low traffic and low demand, which are used for subsequent fuzzy reasoning and power management decisions.

[0093] like Figure 4 As shown, S3 determines the input membership, and the method steps are as follows:

[0094] S3.1.1. Determine the parameters of the Gaussian membership function: the statistical mean of the variable corresponding to the fuzzy set is used as the center position of the Gaussian membership function, and the standard deviation of the fuzzy set data is used to determine the width of the Gaussian membership function;

[0095] S3.1.2, Calculate the triggering strength of the rule: determine the membership of the network traffic and application requirements in the corresponding fuzzy set, and select the minimum value of the input membership as the triggering strength of the rule;

[0096] S3.1.3, Calculate output membership: When the rule is triggered, determine the membership of the output fuzzy set according to the conclusion of the rule;

[0097] Through data collection and analysis, the statistical mean of the variables corresponding to a fuzzy set is obtained, and this mean can be used as the center position of the Gaussian membership function. For example, for the "high traffic" fuzzy set, the average value of the network traffic data transmission rate is μ = 5MB / s, so in the Gaussian membership function In the above equation, μ = 5MB / s means that the membership degree in this fuzzy set is theoretically the highest position;

[0098] Calculate the standard deviation of the data to determine the width of the function. For example, calculate the standard deviation δ for the "high traffic" network traffic data. A smaller δ means that the data is more concentrated near the mean, and the membership function decreases faster; a larger δ means that the data is more dispersed, and the membership function decreases relatively slowly.

[0099] In order to adjust the shape of the membership function, when determining the parameters of the Gaussian membership function, the standard deviation is adjusted by the expected membership change rate to adjust the shape of the membership function;

[0100] The membership change rate describes how fast the membership changes with the variable. In the Gaussian membership function, the change rate is related to the standard deviation. A smaller standard deviation will cause the membership to change quickly around the mean, while a larger standard deviation will cause the membership to change more slowly.

[0101] For example, we expect that within the range of ±1MB / s from the mean, the membership degree decreases from 1 to 0.5, and we also substitute the Gaussian membership function The δ value can be solved. Through such calculation, the standard deviation can be adjusted according to the expected membership change rate to make the shape of the membership function meet the requirements. For example, if the obtained δ value is small, it means that the current membership function decreases too fast and does not meet the expected change rate. It needs to be increased appropriately to make the membership change more smoothly; on the contrary, if the δ value is large, it needs to be reduced to speed up the decline of the membership.

[0102] In fuzzy reasoning, suppose we have a fuzzy rule "If the network traffic is 'high traffic' and the application demand is 'high demand', then the wireless LAN module enters active mode". For each input variable (network traffic and application demand), first determine their membership in their respective fuzzy sets. For example, the membership of network traffic to "high traffic" is calculated by the Gaussian membership function as μ1 = 0.7, and the membership of application demand to "high demand" is μ2 = 0.6. The trigger strength of the rule usually takes the minimum value of the input membership, so the trigger strength of this rule is α = min(μ1,μ2) = 0.6;

[0103] When the rule is triggered, the membership of the output fuzzy set (i.e., active mode) is determined according to the conclusion of the rule. Assuming that no other rules have previously affected the membership of the active mode, the membership of the active mode at this time is equal to the rule triggering strength, i.e., 0.6.

[0104] In order to determine the final output membership, when calculating the output membership, if there are multiple rules that affect the membership of the output fuzzy set, the maximum value method is used to determine the final output membership;

[0105] If there are multiple rules that affect the membership of the output fuzzy set, for example, another rule also leads to a membership of 0.4 for the active mode, these output memberships need to be merged. The maximum value method can be used, that is, the final membership of the active mode is max(0.6,0.4). This means that according to the current input situation and fuzzy rules, the membership of the wireless LAN module entering the active mode is 0.6, and this membership will be used in the subsequent defuzzification process to determine the specific power management parameters.

[0106] In order to convert the membership degree into a specific voltage parameter, the S4 membership degree is converted into a specific voltage parameter, wherein the contribution of each fuzzy set to the power supply parameter is calculated, and the final voltage is obtained according to the voltage conversion formula, wherein the voltage conversion formula is as follows:

[0107]

[0108] Where V is the voltage, V min is the minimum value of the lower limit of the voltage range in all modes, ∑ i μ i is the sum of all memberships, C i is the voltage contribution, and n is the number of memberships.

[0109] Through the fuzzy reasoning process, the membership of each sleep mode in the current state is obtained. For example, after calculation, the membership of the deep sleep mode is μ1 = 0.3, the membership of the light sleep mode is μ2 = 0.5, and the membership of the standby mode is μ3 = 0.2. The minimum value of the voltage lower limit V is determined. min =1.8V, the sum of all memberships Substituting the above calculated values ​​into the formula, we can get The center of gravity method is used to convert the fuzzy sleep mode membership into specific power parameter settings, which can be used to set the power of the wireless LAN module to achieve a suitable power management mode.

[0110] In summary, the working principle of this solution is as follows:

[0111] The power management method of the computer wireless local area network module collects network flow data and application state data through network equipment management software, and uses an autoregressive moving average model to build a flow prediction model to predict the flow usage of the computer. According to the predicted network flow and application requirements, a cluster analysis method is used to dynamically adjust the range of the corresponding fuzzy set, and a corresponding fuzzy rule is established. Then, according to the transmission of flow data, the working mode of the wireless local area network module is dynamically adjusted, the input membership is determined according to the collected network flow data and application requirement data, and the output membership is calculated according to the rule trigger strength. Then, the fuzzy mode membership is converted into a specific voltage parameter by using the center of gravity method, and the power supply is regulated. Through power distribution, the voltage of the wireless local area network module is regulated to a specific value, so as to realize the refined control of the power management of the wireless local area network module and reduce the power consumption.

[0112] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A power management method for a computer wireless local area network module, characterized in that: The following steps are involved: S1. Collect network traffic data and application status data using network device management software, and build a traffic prediction model using an autoregressive moving average model; S2, define network traffic fuzzy set and application requirement fuzzy set, and establish fuzzy rules; S3, determining the input membership according to the collected network traffic data and application demand data, and calculating the output membership according to the rule triggering strength; S4. Use the centroid method to convert the fuzzy mode membership into specific voltage parameters and regulate the power supply.

2. The power management method for a computer wireless local area network module according to claim 1, characterized in that: S1 uses an autoregressive moving average model to build a traffic prediction model, and the steps are as follows: S1.1.1, Data cleaning and processing: Clean the historical network traffic data collected from the wireless LAN module and use the differential method to process the data smoothly; S1.1.

2. Determine model parameters: determine the autoregressive order p of the model using the autocorrelation function, determine the moving average order q using the Yule-Walker equation, and determine the differential order d using the autocorrelation function graph and the partial autocorrelation function graph; S1.1.3, Model fitting: Use the least squares method to estimate the model parameters and test whether the residuals of the model meet the white noise assumption; S1.1.4, Traffic prediction stage: Use the fitted autoregressive moving average model to predict traffic, and the formula is: in, is the autoregressive operator, θ(B) is the moving regression operator, B is the lag operator, and Y t is time series data, ε t is white noise, and d is the difference order.

3. The power management method for a computer wireless local area network module according to claim 2, characterized in that: S1.1.2 uses the autocorrelation function to determine the autoregressive order p of the model, and the formula is: Among them, p k is the autoregressive order of the lag k period, T is the length of the time series, y t is a time series, is the time series y t The mean of , k is the number of lags, y t- k is the time series from k to t.

4. The power management method for a computer wireless local area network module according to claim 2, characterized in that: When fitting the S1.1.3 model, the normality of the residuals was tested using the Shapiro-Wilk method. If the residuals did not conform to the normal distribution, the model parameters were adjusted.

5. The power management method for a computer wireless local area network module according to claim 1, characterized in that: The S2 establishes fuzzy rules, and the method steps are as follows: S2.1.1, Cluster analysis process: Use K-Means clustering algorithm to cluster network traffic data and application demand data; S2.1.2, define the scope of fuzzy sets: analyze each cluster to determine the network state it represents, and determine the scope of fuzzy sets based on the data distribution of cluster centers and members; S2.1.

3. Establish fuzzy rules: Establish fuzzy rules based on the network status represented by the cluster and the actual power management requirements, add device status factors, and refine the rules.

6. The power management method for a computer wireless local area network module according to claim 5, characterized in that: In the cluster analysis process S2.1.1, the elbow rule is used to determine the appropriate number of clusters, measure the relationship between the compactness of clusters and the number of clusters, and find the turning point where the cluster change trend changes from a sharp decline to a gentle decline.

7. The power management method for a computer wireless local area network module according to claim 1, characterized in that: S3 determines the input membership, and the method steps are as follows: S3.1.

1. Determine the parameters of the Gaussian membership function: the statistical mean of the variable corresponding to the fuzzy set is used as the center position of the Gaussian membership function, and the standard deviation of the fuzzy set data is used to determine the width of the Gaussian membership function; S3.1.2, Calculate the triggering strength of the rule: determine the membership of the network traffic and application requirements in the corresponding fuzzy set, and select the minimum value of the input membership as the triggering strength of the rule; S3.1.

3. Calculate output membership: When the rule is triggered, determine the membership of the output fuzzy set based on the conclusion of the rule.

8. The power management method for a computer wireless local area network module according to claim 7, characterized in that: When determining the parameters of the Gaussian membership function in S3.1.1, the standard deviation is adjusted by the expected membership change rate to adjust the shape of the membership function.

9. The power management method for a computer wireless local area network module according to claim 7, characterized in that: When calculating the output membership in S3.1.3, if there are multiple rules that affect the membership of the output fuzzy set, the maximum value method is used to determine the final output membership.

10. The power management method for a computer wireless local area network module according to claim 1, characterized in that: The S4 membership is converted into a specific voltage parameter, wherein the contribution of each fuzzy set to the power supply parameter is calculated, and the final voltage is obtained according to the voltage conversion formula, wherein the voltage conversion formula is as follows: Where V is the voltage, V min is the minimum value of the lower limit of the voltage range in all modes, ∑ i μ i is the sum of all memberships, C i is the voltage contribution, and n is the number of memberships.

Images