WiFi data analysis method based on WiFi indexes

Abstract

The invention discloses a WiFi data analysis method based on a WiFi index. The method comprises the following steps: S1, acquiring WiFi data; s2, cleaning the data; s3, dividing the acquired WiFi indexes into a QOE index and a KPI (Key Performance Indicator) index; s4, establishing a learning device, and learning joint features of the QOE indexes and the KPI indexes; s5, automatically learning the position features of the QOE index values; s6, extracting marginal features of the KPI indexes with conditions; s7, calculating the gradient of the KPI marginal features; s8, calculating a KPI index marginal feature gradient ratio; s9, optimizing the index according to the influence degree of the KPI index on the QOE index; and S10, circulating the steps S4-S9 until all the KPI indexes and the QOE indexes are analyzed. According to the method, WiFi operation and maintenance personnel are guided to optimize the WiFi environment, the influence of expert experience differences is avoided, a static threshold value is not set, and the method adapts to different WiFi environments.

Description

Technical Field

[0001] The present invention relates to the field of network communications, and in particular to a WiFi data analysis method based on WiFi indicators. Background Art

[0002] As an efficient and convenient wireless network communication technology based on the IEEE 802.11 standard, WiFi is gradually becoming an indispensable part of people's daily life and work. With the rapid development of smart homes and the Internet of Things, WiFi has gradually become a key connection technology and the core of connecting various smart devices. While WiFi is widely used, it also puts forward higher requirements for WiFi operation and maintenance efficiency.

[0003] As an important basis for WiFi operation and maintenance, WiFi data directly reflects the actual usage of users. WiFi data can help operation and maintenance personnel quickly locate network problems, so as to take corresponding measures to optimize, ensure service quality and improve user satisfaction.

[0004] Existing WiFi data analysis technologies include manual analysis and comprehensive evaluation. Manual analysis refers to the manual extraction of data by operation and maintenance personnel. Based on their own experience, they set different thresholds for different indicators to judge the quality of WiFi performance data and then take operation and maintenance measures. The comprehensive evaluation method refers to the establishment of a WiFi indicator system in advance, setting segmented thresholds for indicator values, dividing the indicator values ​​into several intervals, assigning a specific score to each interval, determining the interval to which the WiFi performance indicator value actually observed belongs, and then calculating the final score based on the scoring rules of the interval. The scores of multiple indicators are weighted averaged or combined according to certain rules to obtain an overall score. The operation and maintenance personnel judge the user WiFi status based on the overall score.

[0005] Manual analysis of WiFi data is inefficient. The common feature of existing WiFi data analysis technologies is to set thresholds for indicators. The actual observed WiFi indicator values ​​are compared with the thresholds. After calculation, a comprehensive score is obtained. The quality of WiFi is judged by the comprehensive score. However, setting the thresholds completely depends on the subjective experience of the operation and maintenance personnel. The knowledge and operation and maintenance experience of different operation and maintenance personnel vary. It is difficult to accurately judge the quality of WiFi indicator values ​​through experience thresholds. On the other hand, after the threshold is set, it becomes a static value, which is difficult to adapt to various different WiFi environments. Summary of the invention

[0006] In order to solve the above technical problems, the present invention proposes a WiFi data analysis method based on WiFi indicators, which overcomes the shortcomings of the existing WiFi data analysis technology that relies on low efficiency of manual analysis, relies on expert experience, has difficulty in accurately judging the quality of Wifi indicator values ​​through empirical thresholds, and is difficult to adapt to different WiFi environments by setting static thresholds.

[0007] A WiFi data analysis method based on WiFi indicators includes the following steps:

[0008] S1: Obtain WiFi data from an external data system;

[0009] S2: Clean the data with missing values, duplications, and illogical data;

[0010] S3: Divide the acquired WiFi indicators into QOE indicators and KPI indicators;

[0011] S4: Establish a learner to automatically learn the joint features of QOE indicators and KPI indicators from data;

[0012] S5: The learner automatically learns the positional features of the QOE indicator values ​​from the data;

[0013] S6: Extract marginal features of conditional KPI indicators;

[0014] The condition is that the QOE index value ≤ the position characteristic value of the QOE index value;

[0015] S7: Calculate the gradient of the marginal characteristics of the KPI indicator;

[0016] S8: Calculate the ratio of the marginal characteristic gradient of the KPI indicator;

[0017] S9: Optimize the indicators according to the impact of KPI indicators on QOE indicators;

[0018] S10: looping steps S4 to S9 until the analysis of all KPI indicators and QOE indicators is completed.

[0019] Furthermore, a WiFi data analysis method based on WiFi indicators is provided, wherein the QOE indicator reflects the user's subjective acceptability of the application or service as a whole, and the KPI indicator is a key network performance indicator.

[0020] Furthermore, in a WiFi data analysis method based on WiFi indicators, step S4 includes the following sub-steps:

[0021] When it is a continuous indicator, the learner establishes a joint feature learning function for the continuous indicator, and the formula is:

[0022]

[0023] Among them, F(x, y) represents the joint eigenvalue, x 0 Represents the upper limit of the KPI index, y 0 represents the upper limit of the QOE index interval, X represents the KPI index value, Y represents the QOE index value, P represents the probability of meeting the conditions in the brackets, f(x, y) represents the probability density function of the joint distribution of x and y, and x and y represent the input function variables;

[0024] When it is a discrete indicator, the learner establishes a joint feature learning function for the discrete indicator, and the formula is:

[0025]

[0026] Among them, F(x, y) represents the joint eigenvalue, x 0 Represents the upper limit of the KPI indicator, y 0 represents the upper limit of the QOE index, X represents the KPI index value, Y represents the QOE index value, and p(x, y) represents the probability mass function of the joint distribution of x and y.

[0027] Furthermore, in a WiFi data analysis method based on WiFi indicators, step S5 includes the following sub-steps:

[0028] When the amount of data is an odd number, the learner establishes a learning function for the positional features of the odd QOE indicator value, and the formula is:

[0029]

[0030] Among them, t represents the position characteristic value of the QOE indicator value, sort represents the ascending sort of the indicator value, Y represents the QOE indicator value, and n represents the amount of indicator value data;

[0031] When the amount of data is an even number, the learner establishes a learning function for the positional features of the even-numbered QOE indicator value, and the formula is:

[0032]

[0033] Among them, t represents the position characteristic value of the QOE indicator value, sort represents the ascending sort of the indicator value, Y represents the QOE indicator value, n represents the amount of indicator value data, and int represents the rounding operation of the number.

[0034] Furthermore, in a WiFi data analysis method based on WiFi indicators, step S6 includes the following sub-steps:

[0035] When it is a continuous indicator, the marginal features of the conditional KPI indicator are extracted, and the formula is:

[0036]

[0037] Among them, F(x) represents the marginal eigenvalue of the conditional KPI indicator, F(x, y) represents the joint eigenvalue, t represents the positional eigenvalue of the QOE indicator value, and f(x, t) represents the joint distribution probability density function of x and y when y=t is given;

[0038] When it is a discrete indicator, the marginal characteristics of the conditional KPI indicator are extracted, and the formula is:

[0039]

[0040] Among them, F(x) represents the marginal eigenvalue of the conditional KPI indicator, F(x, y) represents the joint eigenvalue, t represents the positional eigenvalue of the QOE indicator value, and p(x, t) represents the probability mass function of the joint distribution of x and y when y=t is given.

[0041] Furthermore, a WiFi data analysis method based on WiFi indicators, said S7 comprises the following sub-steps:

[0042] S71: Calculate the gradient of the marginal characteristic of the KPI indicator. The formula is:

[0043]

[0044] Among them, F′(x) represents the gradient of the marginal characteristic of the KPI indicator, Δx represents the increment of x, and F(x) represents the marginal characteristic value of the conditional KPI indicator.

[0045] Furthermore, a WiFi data analysis method based on WiFi indicators, said S8 comprises the following sub-steps:

[0046] S81: Calculate the ratio of the KPI indicator marginal characteristic gradient, the formula is:

[0047]

[0048] Among them, r represents the ratio of the marginal characteristic gradients of the two KPI indicators, F′(X1) represents the marginal characteristic gradient of one of the two KPI indicators being compared, and F′(X2) represents the marginal characteristic gradient of the other of the two KPI indicators being compared.

[0049] Furthermore, a WiFi data analysis method based on WiFi indicators, the S9 includes the following sub-steps:

[0050] If r>1, the impact of KPI X1 on QOE is greater than the impact of KPI X2 on QOE, so KPI X1 should be optimized first.

[0051] If r<1, the impact of KPI indicator X2 on the QOE indicator is greater than the impact of KPI indicator X1 on the QOE indicator, and KPI indicator X2 is optimized first.

[0052] The beneficial effects of the present invention are as follows: a WiFi data analysis method based on WiFi indicators, which respectively realizes obtaining WiFi data, cleaning WiFi data, and establishing a learner, automatically learning the joint features between QOE indicators and KPI indicators from the data, calculating the marginal feature gradient of the KPI indicator, evaluating the influence of the KPI indicator on the QOE indicator, and determining the WiFi indicators to be improved first, thereby guiding WiFi operation and maintenance personnel to optimize the WiFi environment, is not affected by differences in expert experience, does not set static thresholds, and adapts to different WiFi environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flow chart of a WiFi data analysis method based on WiFi indicators. DETAILED DESCRIPTION

[0054] The present invention is further described below, but the protection scope of the present invention is not limited to the following description.

[0055] As attached Figure 1 As shown, a WiFi data analysis method based on WiFi indicators includes the following steps:

[0056] S1: Obtain WiFi data from an external data system;

[0057] S2: Clean the data with missing values, duplications, and illogical data;

[0058] S3: Divide the acquired WiFi indicators into QOE indicators and KPI indicators;

[0059] S4: Establish a learner to automatically learn the joint features of QOE indicators and KPI indicators from data;

[0060] S5: The learner automatically learns the positional features of the QOE indicator values ​​from the data;

[0061] S6: Extract marginal features of conditional KPI indicators;

[0062] The condition is that the QOE index value ≤ the position characteristic value of the QOE index value;

[0063] S7: Calculate the gradient of the marginal characteristics of the KPI indicator;

[0064] S8: Calculate the ratio of the marginal characteristic gradient of the KPI indicator;

[0065] S9: Optimize the indicators according to the impact of KPI indicators on QOE indicators;

[0066] S10: looping steps S4 to S9 until the analysis of all KPI indicators and QOE indicators is completed.

[0067] Furthermore, a WiFi data analysis method based on WiFi indicators is provided, wherein the QOE indicator reflects the user's subjective acceptability of the application or service as a whole, and the KPI indicator is a key network performance indicator.

[0068] Furthermore, in a WiFi data analysis method based on WiFi indicators, step S4 includes the following sub-steps:

[0069] When it is a continuous indicator, the learner establishes a joint feature learning function for the continuous indicator, and the formula is:

[0070]

[0071] Among them, F(x, y) represents the joint eigenvalue, x 0 Represents the upper limit of the KPI indicator, y 0 represents the upper limit of the QOE index interval, X represents the KPI index value, Y represents the QOE index value, P represents the probability of meeting the conditions in the brackets, f(x, y) represents the probability density function of the joint distribution of x and y, and x and y represent the input function variables;

[0072] When it is a discrete indicator, the learner establishes a joint feature learning function for the discrete indicator, and the formula is:

[0073]

[0074] Among them, F(x, y) represents the joint eigenvalue, x 0 Represents the upper limit of the KPI indicator, y 0 represents the upper limit of the QOE index, X represents the KPI index value, Y represents the QOE index value, and p(x, y) represents the probability mass function of the joint distribution of x and y.

[0075] Furthermore, in a WiFi data analysis method based on WiFi indicators, step S5 includes the following sub-steps:

[0076] When the amount of data is an odd number, the learner establishes a learning function for the positional features of the odd QOE indicator value, and the formula is:

[0077]

[0078] Among them, t represents the position characteristic value of the QOE indicator value, sort represents the ascending sort of the indicator value, Y represents the QOE indicator value, and n represents the amount of indicator value data;

[0079] When the amount of data is an even number, the learner establishes a learning function for the positional features of the even-numbered QOE indicator value, and the formula is:

[0080]

[0081] Among them, t represents the position characteristic value of the QOE indicator value, sort represents the ascending sort of the indicator value, Y represents the QOE indicator value, n represents the amount of indicator value data, and int represents the rounding operation of the number.

[0082] Furthermore, in a WiFi data analysis method based on WiFi indicators, step S6 includes the following sub-steps:

[0083] When it is a continuous indicator, the marginal features of the conditional KPI indicator are extracted, and the formula is:

[0084]

[0085] Among them, F(x) represents the marginal eigenvalue of the conditional KPI indicator, F(x, y) represents the joint eigenvalue, t represents the positional eigenvalue of the QOE indicator value, and f(x, t) represents the joint distribution probability density function of x and y when y=t is given;

[0086] When it is a discrete indicator, the marginal characteristics of the conditional KPI indicator are extracted, and the formula is:

[0087]

[0088] Among them, F(x) represents the marginal eigenvalue of the conditional KPI indicator, F(x, y) represents the joint eigenvalue, t represents the positional eigenvalue of the QOE indicator value, and p(x, t) represents the probability mass function of the joint distribution of x and y when y=t is given.

[0089] Furthermore, a WiFi data analysis method based on WiFi indicators, said S7 comprises the following sub-steps:

[0090] S71: Calculate the gradient of the marginal characteristic of the KPI indicator. The formula is:

[0091]

[0092] Among them, F′(x) represents the gradient of the marginal characteristic of the KPI indicator, Δx represents the increment of x, and F(x) represents the marginal characteristic value of the conditional KPI indicator.

[0093] Furthermore, a WiFi data analysis method based on WiFi indicators, said S8 comprises the following sub-steps:

[0094] S81: Calculate the ratio of the KPI indicator marginal characteristic gradient, the formula is:

[0095]

[0096] Among them, r represents the ratio of the marginal characteristic gradients of the two KPI indicators, F′(X1) represents the marginal characteristic gradient of one of the two KPI indicators being compared, and F′(X2) represents the marginal characteristic gradient of the other of the two KPI indicators being compared.

[0097] Furthermore, a WiFi data analysis method based on WiFi indicators, the S9 includes the following sub-steps:

[0098] If r>1, the impact of KPI X1 on QOE is greater than the impact of KPI X2 on QOE, so KPI X1 should be optimized first.

[0099] If r<1, the impact of KPI indicator X2 on the QOE indicator is greater than the impact of KPI indicator X1 on the QOE indicator, and KPI indicator X2 is optimized first.

[0100] A WiFi data analysis method based on WiFi indicators. The QOE indicator takes the download rate as an example, and the KPI indicator takes the WiFi signal strength and the number of WiFi interference as examples. The steps are as follows:

[0101] Step 1: Calculate t=100 through S1~S5;

[0102] Step 2: Calculate through S6~S7 that the WiFi signal strength is equal to -67db, the marginal feature gradient is 0.0331, and when the number of WiFi interference is equal to 3, the marginal feature gradient is equal to 0.0173;

[0103] Step 3: When the WiFi signal strength is -67 db and the number of WiFi interferences is 3, the ratio of the marginal feature gradient is r=0.0331 / 0.0173=1.9133.

[0104] The X1 indicator is the WiFi signal strength, and the X2 indicator is the amount of WiFi interference;

[0105] Step 4: When the WiFi signal strength is equal to -67db and the number of WiFi interferences is equal to 3, it is determined through S9 that the impact of the WiFi signal strength on the download rate is greater than the impact of the number of WiFi interferences on the download rate, and the WiFi signal strength is optimized first.

[0106] This solution uses a WiFi data analysis method based on WiFi indicators to obtain WiFi data, clean WiFi data, and establish a learner. It automatically learns the joint features between QOE indicators and KPI indicators from the data, calculates the marginal feature gradient of KPI indicators, evaluates the impact of KPI indicators on QOE indicators, and determines the WiFi indicators that need to be improved first, thereby guiding WiFi operation and maintenance personnel to optimize the WiFi environment. It is not affected by differences in expert experience, does not set static thresholds, and can adapt to different WiFi environments.

[0107] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of 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 to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A WiFi data analysis method based on WiFi indicators, characterized in that: The following steps are involved: S1: Obtain WiFi data from an external data system; S2: Clean the data with missing values, duplications, and illogical data; S3: Divide the acquired WiFi indicators into QOE indicators and KPI indicators; S4: Establish a learner to automatically learn the joint features of QOE indicators and KPI indicators from data; S5: The learner automatically learns the positional features of the QOE indicator values ​​from the data; S6: Extract marginal features of conditional KPI indicators; The condition is that the QOE index value is less than or equal to the position characteristic value of the QOE index value; S7: Calculate the gradient of the marginal characteristics of the KPI indicator; S8: Calculate the ratio of the marginal characteristic gradient of the KPI indicator; S9: Optimize the indicators according to the impact of KPI indicators on QOE indicators; S10: looping steps S4 to S9 until the analysis of all KPI indicators and QOE indicators is completed.

2. The WiFi data analysis method based on WiFi indicators according to claim 1, characterized in that: The QOE indicator reflects the user's subjective acceptability of the application or service as a whole, and the KPI indicator is a key network performance indicator.

3. The WiFi data analysis method based on WiFi indicators according to claim 1, characterized in that: Step S4 includes the following sub-steps: When it is a continuous indicator, the learner establishes a joint feature learning function for the continuous indicator, and the formula is: Among them, F(x, y) represents the joint eigenvalue, x0 represents the upper limit of the interval of the KPI indicator, y0 represents the upper limit of the interval of the QOE indicator, X represents the KPI indicator value, Y represents the QOE indicator value, P represents the probability of meeting the conditions in the brackets, f(.xy) represents the probability density function of the joint distribution of .x and y, and .x and y represent the input function variables; When it is a discrete indicator, the learner establishes a joint feature learning function for the discrete indicator, and the formula is: Among them, F(x, y) represents the joint eigenvalue, x0 represents the upper limit of the interval of the KPI indicator, y0 represents the upper limit of the interval of the QOE indicator, X represents the KPI indicator value, Y represents the QOE indicator value, and p(xy) represents the probability mass function of the joint distribution of x and y.

4. The WiFi data analysis method based on WiFi indicators according to claim 1, characterized in that: The step S5 comprises the following sub-steps: When the amount of data is an odd number, the learner establishes a learning function for the positional features of the odd QOE indicator value, and the formula is: Among them, t represents the position characteristic value of the QOE indicator value, sort represents the ascending sort of the indicator value, Y represents the QOE indicator value, and n represents the amount of indicator value data; When the amount of data is an even number, the learner establishes a learning function for the positional features of even-numbered QOE indicator values, and the formula is: Among them, t represents the position characteristic value of the QOE indicator value, sort represents the ascending sort of the indicator value, Y represents the QOE indicator value, n represents the amount of indicator value data, and int represents the rounding operation of the number.

5. The WiFi data analysis method based on WiFi indicators according to claim 1, characterized in that: The step S6 comprises the following sub-steps: When it is a continuous indicator, the marginal features of the conditional KPI indicator are extracted, and the formula is: Among them, F(x) represents the marginal eigenvalue of the conditional KPI indicator, F(x, y) represents the joint eigenvalue, t represents the positional eigenvalue of the QOE indicator value, and f(x, t) represents the joint distribution probability density function of x and y when y=t is given; When it is a discrete indicator, the marginal features of the conditional KPI indicator are extracted, and the formula is: Among them, F(x) represents the marginal eigenvalue of the conditional KPI indicator, F(xy) represents the joint eigenvalue, t represents the positional eigenvalue of the QOE indicator value, and p(x, t) represents the probability mass function of the joint distribution of x and y when y=t is given.

6. A WiFi data analysis method based on WiFi indicators according to claim 1, characterized in that: The S7 comprises the following sub-steps: S71: Calculate the gradient of the marginal characteristic of the KPI indicator. The formula is: Among them, F′(.x) represents the gradient of the marginal characteristic of the KPI indicator, Δx represents the increment of x, and F(x) represents the marginal characteristic value of the conditional KPI indicator.

7. The WiFi data analysis method based on WiFi indicators according to claim 1, characterized in that: The S8 comprises the following sub-steps: S81: Calculate the ratio of the KPI indicator marginal characteristic gradient, the formula is: Among them, r represents two KPs I The ratio of the marginal characteristic gradients of the indicators, F′(X1) represents the marginal characteristic gradient of one of the two KPI indicators being compared, and F′(X2) represents the marginal characteristic gradient of the other of the two KPI indicators being compared.

8. The WiFi data analysis method based on WiFi indicators according to claim 1, characterized in that: The S9 comprises the following sub-steps: If r>1, the impact of KPI indicator x1 on QOE indicator is greater than the impact of KPI indicator X2 on QOE indicator, and KPI indicator X1 is optimized first; If r<1, the impact of KPI indicator X2 on the QOE indicator is greater than the impact of KPI indicator X1 on the QOE indicator, and KPI indicator X2 is optimized first.

Images