KR101611355B1 - Automatic measurement apparatus and method for mobile communication system applying an adaptive sampling algorithm - Google Patents

Abstract

The present invention relates to a device and method for automatically measuring a mobile communications system having an adaptive sampling scheme applied thereto. Even when data in one frequency band are not predicted by utilizing measurement data of another frequency band based on correlation between measurement data or a lot test is not performed through sampling inspection, quality of an overall product may be ensured. The method according to the present invention comprises the steps of: confirming data with respect to each item of a frequency band selectable in a mobile communications system, and constructing the data as a database; constructing a regression model for predicting a measurement value of the items in a non-measured frequency band by using a measurement value of an item in a measured frequency band; estimating an aposteriori distribution of the item result in a frequency band by applying the Bayesian analysis scheme to the regression model; and measuring an item representative value in the frequency band with respect to a sample in the mobile communications system to be inspected, and predicting the item value in a remaining frequency band through the regression model when the item representative value is suitable for fair quality data.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic measurement apparatus and method for an automatic measurement system for a mobile communication system employing an adaptive sampling technique,

The present invention relates to a method and apparatus for estimating data in different frequency bands using measurement data of one frequency band based on correlation between measurement data or an adaptive type And an apparatus and method for automatically measuring a mobile communication system to which a sampling technique is applied.

Generally, a mobile communication system includes a mobile communication terminal, a mobile communication relay device, and the like.

Interference Cancellation System (ICS) The mobile communication relay system is a relay system that detects and removes RF interference signals generated by feedback between antennas in real time using digital signal processing technology.

In particular, in the case of a WCDMA band selective ICS repeater, a broad bandwidth is input in hardware to correspond to different frequency bands for each communication company, and a part thereof is selected and used. In this case, the manufacturer is not sure which bandwidth to use in the field, so it should be tested against all cases.

At present, the testing method of the mobile communication relay system takes a lot of time in the test process because the operator operates the measuring device by measuring the test items by the test items. Especially, in the case of WCDMA Band Selective ICS repeater, since the same experiment is repeatedly performed for each frequency band, an increase in measurement cost and a decrease in productivity are a great impediment to competition with low-priced Chinese analog products.

A problem to be solved by the present invention is to provide a method of estimating the quality of an entire product even if the data of another frequency band is predicted by utilizing the measurement data of one frequency band based on the correlation between measurement data, And an apparatus and method for automatically measuring a mobile communication system using an adaptive sampling technique.

A method for automatically measuring a mobile communication system using an adaptive sampling technique according to the present invention includes the steps of constructing a database by checking data on each item of a frequency band that can be selected in a mobile communication system; Constructing a regression model for predicting a measured value of items of a non-measured frequency band as a measure of an item of the measured frequency band; Estimating a posterior distribution of item results of a frequency band by applying Bayesian analysis to the regression model; And estimating a representative value of each of the frequency bands for the sample among the mobile communication systems to be inspected and predicting the item values of the remaining frequency bands by a regression model when the representative values are suitable for the good data.

Preferably, the step of constructing the database further includes a step of confirming the correlation for each item and classifying the items for which correlation has been confirmed into a set.

Preferably, the step of constructing the regression model comprises the steps of: measuring fitness of each frequency band for the regression model; And estimating a predictive power of the regression model.

Preferably, estimating the posterior distribution comprises: calculating a prior distribution; Calculating a likelihood function; Confirming a posterior distribution by applying a Bayesian analysis technique using a prior distribution and a likelihood function; And estimating the expected value of the parameter from the posterior distribution.

A mobile communication system automatic measuring apparatus to which an adaptive sampling technique according to the present invention is applied includes a frequency band data management unit (10) for checking data on each item of a frequency band in a mobile communication system and constructing the data as a database; A regression model providing unit 20 for providing a regression model for predicting a measured value of items of a non-measured frequency band as a measured value of an item of the measured frequency band; A frequency band analyzer 30 for estimating a posterior distribution of the frequency band item results by applying Bayesian analysis to the regression model provided by the regression modeling unit 20; And a measurement data analysis unit (40) for measuring a representative value of each frequency band of the sample in the mobile communication system and predicting the item value of the remaining frequency band by a regression model when the normal distribution is satisfied.

Preferably, the frequency band data management unit 10 identifies correlations for each item, and classifies the correlated items as a set.

Preferably, the regression modeling unit 20 includes a fitness check unit 21 for measuring fitness with respect to the regression model presented in the mobile communication system; And a prediction power confirmation unit 22 for confirming a prediction power of the regression model of the mobile communication system.

Preferably, the frequency band analyzing unit 30 includes a pre-distribution calculating unit 31 for confirming a pre-distribution from the regression model, and a pre-distribution calculating unit 31 for calculating a likelihood function using the frequency band of the mobile communication system stored in the frequency- A posterior distribution confirmation unit 32 for confirming the posterior distribution by combining the prior distribution and the likelihood function, an expected value estimation unit 33 for estimating the expected value of the parameter in the posterior distribution by the posterior distribution confirmation unit 32, And a control unit.

The apparatus for automatically measuring a mobile communication system to which the adaptive sampling technique according to the present invention is applied is characterized in that it uses measurement data of one frequency band based on the correlation between measurement data in a mobile communication system of a mobile communication relay apparatus, It is possible to guarantee the quality of the whole product without predicting the data of the frequency band or performing the full inspection by sampling inspection.

1 is a block diagram of a mobile communication system automatic measuring apparatus to which an adaptive sampling technique according to the present invention is applied.
FIG. 2 is a graph illustrating correlation of frequency bands in an apparatus for automatically measuring a mobile communication system using an adaptive sampling technique according to the present invention.
FIG. 3 is a flowchart illustrating an analysis process of the mobile communication system automatic measuring apparatus to which the adaptive sampling technique according to the present invention is applied.
FIG. 4 illustrates an overall structure of a mobile communication system automatic measuring apparatus to which an adaptive sampling technique according to the present invention is applied.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a mobile communication system automatic measuring apparatus to which an adaptive sampling technique according to the present invention is applied will be described in detail with reference to the accompanying drawings.

1, an apparatus for automatically measuring a mobile communication system using an adaptive sampling technique according to the present invention includes a frequency band data management unit 10, a regression model providing unit 20, a frequency band analyzing unit 30, And a data analyzing unit 40. The mobile communication system has a plurality of frequency bands, and is a system for confirming good products of a non-measured frequency band. At this time, the mobile communication system includes a mobile communication relay device, a mobile communication terminal, and the like.

The frequency band data management unit 10 identifies the respective items of the frequency band in the mobile communication system to be established and constructs the items as a database. At this time, if the correlation is confirmed for each item, the items are classified into sets. For example, as shown in FIG. 2, 450 samples are extracted from a mobile communication system having three frequency bands and a matrix plot (matrix scatter diagram) is obtained, which shows a high correlation. At this time, if the confidence interval is about 95%, the item is classified as a set.

The regression modeling unit 20 provides a regression model for predicting the measured values of the items of the unmeasured frequency band as the measured values of the items of the measured frequency band. For example, if a dependent variable is predicted or estimated by providing a simple regression model, one independent variable (measured value) and one dependent variable, two independent variables (measured value), and a multivariate regression model To predict or estimate the dependent variable. An example of the multivariate regression model used in the present invention is expressed as Equation (1).

At this time, Y, ε are represented by n × p, X is n × (q + 1), and B is represented by matrix form of (q + 1) × p. Where Y is the dependent variable for the item in the unmeasured frequency band and X is the independent variable for the item in the measured frequency band.

When applied to mobile communication system data having three frequency bands, p = 2, q = 1, and is expressed by Equation (2).

Accordingly, y in the first column of the matrix is expressed by Equation (3), and y in the second column of the matrix is expressed by Equation (4).

의 추정식은 [수학식 5]와 같고,

의 추정식과

matrix 형태는 [수학식 6]으로 표현된다. 이때, 첫 번째 열의 성분이 y1의 회귀모형으로 사용되고, 두 번째 열의 경우 y2의 회귀모형으로 사용된다.On the other hand, the least square estimator for the regression model of (y1, y2) for (x)

Is expressed by Equation (5)

And

The matrix form is expressed by Equation (6). In this case, the components of the first column are used as a regression model of y1, and the second column is used as a regression model of y2.

Using the multivariate regression model, we can obtain the results as shown in [Table 1].

Regression model (Y variable)

　

　

RMSEP Multivariate regression (BAND1) -0.5211 0.99334 0.06489 Multivariate regression (BAND3) 1.12937 0.97446 0.06352

The regression modeling unit 20 includes a fitness check unit 21 of a regression model and a prediction power verification unit 22 of a regression model.

The fitness check unit 21 measures the fitness of the regression model for the mobile communication system. If the p-value is less than 0.05 at α = 0.05, then the independent variable is statistically significant. In addition, the variance analysis of the regression model shows that the regression model is statistically similar if the p-value is less than 0.05 at α = 0.05. In addition, if the ratio of variation explained by the regression model as shown in Equation (7), that is, the coefficient of determination (R ² ) is 0.8 or more, it can be judged that the regression model is appropriate.

Where 0 ≤ R ² ≤1, SS _T represents the sum of squares and SS _R represents the regression sum of squares.

Also, when the number of independent variables is two or more as in Equation (8), the fitness can be evaluated using the modified R ² (adjusted decision coefficient).

≤1, n은 표본개수, k는 설명변수의 개수를 나타낸다.At this time,

≤ 1, n is the number of samples, and k is the number of explanatory variables.

The predictive power checking unit 22 confirms the predictive power of the regression model of the mobile communication system. The predictive power checking unit 22 measures the average prediction error [RMSEP (Root Mean Squared Error of Prediction)] as shown in Equation (9). At this time, it can be estimated that the smaller the average prediction error value, the better the prediction performance.

는 실측값을 의미하고,

는 예측값을 의미한다.At this time, explanation of each variable is necessary

Means an actual value,

Is a predicted value.

The frequency band of BAND2 is measured using the multivariate regression model according to the present invention, and the predictive power for the frequency bands BAND1 and BAND3 can be obtained as shown in [Table 2].

Dependent variable R ² Adjusted R ² RMSEP BAND1 0.9816 0.9815 0.06489 BAND3 0.9772 0.9771 0.06352

The frequency band analyzing unit 30 estimates the posterior distribution of the frequency band item results by applying Bayesian analysis to the regression model provided by the regression model providing unit 20. The frequency band analyzing unit 30 includes a pre-distribution calculating unit 31, a post-distribution checking unit 32, and an expected value estimating unit 33. [

The frequency band analyzing unit 30 analyzes the frequency band results by applying a Bayesian analysis method. The Bayesian analysis method is a method of estimating from a prior distribution and an observed result Observed This is one of the statistical methods to confirm the posterior distribution of the parameters.

Therefore, in order to apply Bayesian analysis method, first distribution and likelihood function should be defined. Here, the pre-distribution uses the pre-distribution calculated using the above-described regression model.

The pre-distribution in the pre-distribution calculating unit 31 can be divided into cases in which the pre-information is known and those in which the pre-information is unknown. When the dictionary information is known, the conjugate dictionary distribution and the information dictionary distribution are used in the present invention. The information dictionary distribution can determine distribution and parameters as past data with a prior distribution using past data. If you do not know the dictionary information, you use the no-information dictionary distribution, which means a dictionary distribution that will perform only minimal tasks for the post-distribution.

Next, in the case of the likelihood function, as described above, the posterior distribution confirmation unit 32 calculates using the following equation (10).

The posterior distribution confirmation unit 32 calculates the likelihood function by applying the result of the frequency band of the mobile communication system stored in the frequency band data management unit 10 to the probability model, generates a posterior distribution by combining the preliminary distribution and the likelihood function . The posterior distribution for the parameter < RTI ID = 0.0 ># is < / RTI >

The expected value estimating unit 33 estimates the expected value of the parameter in the posterior distribution by the posterior distribution confirming unit 32. [ For example, the parameters are estimated from the posterior distribution using the Markov Chain monte Carlo (MCMC) method.

Specifically, the expected value estimating unit 33 estimates the posterior distribution of the parameter by the posterior distribution confirming unit 32 as shown in Equation (11) and estimates the expected value of the parameter in the posterior distribution using Equation (12) do.

The measurement data analysis unit 40 determines one of the values corresponding to one of the groups having a high correlation among the sets of items of the frequency band for the sample among the mobile communication systems as a representative value and measures it by Bayesian regression analysis The item values of the remaining frequency bands are predicted through regression analysis.

The analysis process of the mobile communication system automatic measuring apparatus to which the adaptive sampling technique is applied will be described in detail with reference to FIG.

First, before entering the analysis process, it is determined whether the model of the product to be measured is a new product or an old product. If the new product is a new product, the data of the product is extracted through inspection for all the items, A pre-processing process is included to perform correlation analysis if the evaluation criteria are satisfied after sampling inspection for approximately 5 samples.

All the items of the frequency band that can be selected in the mobile communication system are examined and the results of the good data and the bad data are extracted (step S101), and the correlations between the items are analyzed by analyzing the correlations of the extracted data items (Step S102). At this time, a high correlation can be confirmed for each frequency band according to the result of obtaining the matrix plot (matrix scatter diagram). At this time, items for which correlation has been proved for each frequency band are classified into sets (step S103).

Thereafter, a regression model is constructed to measure one item among the set-proven items correlated in the mobile communication system and to predict other items (step S104). In order to analyze the collected data, the correlation is analyzed by a multivariate regression model such as [Equation 1] and [Equation 2], and the prediction error is confirmed by checking the average prediction error of the regression model as shown in [Equation 9] do.

Next, in order to estimate the expected value using the Bayesian analysis technique, a pre-distribution according to the item is derived in the regression model (step S105), and the posterior distribution is derived by combining the pre-distribution with the likelihood function calculated from the regression model ). At this time, the likelihood function is calculated according to Equation (10), and the MCMC method is used to estimate the expected value from the posterior distribution.

Then, a representative value corresponding to the same set of items in n samples is measured (step S107), and it is determined whether the measured item is located within a predetermined standard of the confidence interval (step 108).

If it is located within the predetermined standard, the data of the remaining items classified as the set are predicted through regression analysis (step S109). At this time, the confidence interval is about 95%, and if the confidence interval of the predicted item is out of the predetermined standard, the all / all items are inspected (step S110).

FIG. 4 illustrates an overall structure of a mobile communication system automatic measuring apparatus to which the adaptive sampling technique of the present invention is applied.

The mobile communication system automatic measuring apparatus of the present invention is connected to a network analyzer, a signal generator, a spectrum analyzer, etc., and automatically inspects the mobile communication system without interrogating the mobile communication system. In addition, the accuracy of the inspection can be improved by reflecting changes in physical properties of the manufacturing lot in real time in the inspection, and the repeater and the measuring instrument can be simultaneously controlled, thereby realizing an unmanned inspection system.

It will be understood by those skilled in the art that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive. The scope of the present invention is defined by the appended claims rather than the detailed description and all changes or modifications derived from the meaning and scope of the claims and their equivalents are to be construed as being included within the scope of the present invention do.

10: frequency band data management unit 20:
21: fitness check part 22: prediction power check part
30: frequency band analyzing unit 31: pre-distribution calculating unit
32: posterior distribution confirming unit 33: expected value estimating unit
40: Measurement data analysis section

Claims (8)

Checking data on each item of a frequency band selectable in the mobile communication system and constructing the data as a database;
Constructing a regression model for predicting a measured value of items of a non-measured frequency band as a measure of an item of the measured frequency band;
Estimating a posterior distribution of item results of a frequency band by applying Bayesian analysis to the regression model; And
Measuring a representative value of each frequency band for each sample in the mobile communication system to be inspected and estimating an item value of the remaining frequency band using a regression model if the representative value is suitable for good data; A method for automatic measurement of a mobile communication system using the method.
The method as claimed in claim 1, wherein the step of constructing the database further comprises the step of checking correlations for each item and classifying the correlated items as a set. The mobile communication system according to claim 1, How to measure.
The method of claim 1, wherein the step of constructing the regression model comprises: measuring fitness of each frequency band of the regression model; And estimating a predictive power of the regression model based on the adaptive sampling method.
2. The method of claim 1, wherein estimating the posterior distribution comprises: calculating a prior distribution;
Calculating a likelihood function;
Confirming the posterior distribution by applying a Bayesian analysis technique using the pre-distribution and the likelihood function;
Estimating an expected value of a parameter from the posterior distribution; and estimating an expected value of a parameter from the posterior distribution.
A frequency band data management unit 10 for confirming data of each item of the frequency band in the mobile communication system and constructing the data as a database;
A regression model providing unit 20 for providing a regression model for predicting a measured value of items of a non-measured frequency band as a measured value of an item of the measured frequency band;
A frequency band analyzer 30 for applying Bayesian analysis to the regression model provided by the regression model providing unit 20 to estimate a posterior distribution of item results of the frequency band;
And a measurement data analysis unit (40) for measuring a representative value of each frequency band for the sample in the mobile communication system and for predicting the item value of the remaining frequency band using the regression model if the sample is suitable for good data Automatic Measurement System of Mobile Communication System Using Adaptive Sampling Technique.
The apparatus of claim 5, wherein the frequency band data management unit (10) identifies correlations for each item and classifies the correlated items as a set.
The system according to claim 5, wherein the regression modeling unit (20) includes a fitness check unit (21) for measuring a fitness of the regression model presented in the mobile communication system;
And a predictive power checking unit (22) for checking the predictive power of the regression model of the mobile communication system.
The system according to claim 5, wherein the frequency band analyzer (30) comprises a pre-distribution calculating unit (31) for confirming a pre-distribution from the regression model,
A posterior distribution confirmation unit 32 for calculating a likelihood function using the frequency band of the mobile communication system stored in the frequency band data management unit 10 and combining the preliminary distribution with the likelihood function to confirm the posterior distribution, ,
And an expected value estimating unit (33) for estimating the expected value of the parameter in the posterior distribution by the posterior distribution confirming unit (32).

Images