JP6466864B2 - Analysis apparatus and analysis method - Google Patents

Description

  The present invention relates to an analysis apparatus and an analysis method.

  Mobile communication is in a situation where quality degradation is more prominent due to various events due to diversification of user environments and diversification of usage methods. Therefore, in order to improve the quality of mobile communication services, it is important to understand the quality degradation factors.

JP2015-115672A

  In order to take appropriate actions for the purpose of improving the quality of mobile communication services, it is necessary to capture not only the fact that the quality has deteriorated but also the events that cause the quality deterioration.

  In existing research, there is a technology that estimates which of the nodes, links, and network sections in the network is the cause of quality degradation, but the event that causes one factor affects multiple locations. However, it is difficult to fundamentally solve the problem unless the individual causes are found and dealt with, and the major causes affecting the plurality of locations are found and not dealt with. Therefore, it is required not only to capture individual deterioration occurrence locations, but also to be able to estimate events that are fundamental factors that have an influence on the deterioration.

  For that purpose, it is considered effective to specify the range of the degradation influence in which the degradation factors are common.

  The present invention has been made in view of the above points, and an object of the present invention is to make it possible to specify a range of deterioration influence that causes a common deterioration factor of communication quality.

  Therefore, in order to solve the above-described problem, the analysis apparatus includes one or more time widths with reference to a storage unit that stores, in time series, the degree of deterioration of communication quality for each time width for each of a plurality of attributes related to communication. For every two durations, for all two attribute sets, the calculation unit that calculates the degree of similarity of the degree of deterioration in the duration, and among the calculated similarities, the duration is common, and at least the attribute An aggregation unit that aggregates two or more similarities that are partly in common, and an output unit that outputs a duration related to the aggregated similarity and a set of attributes;

  It is possible to specify a range of degradation influence that has common causes of degradation of communication quality.

It is a figure which shows the hardware structural example of the analyzer in embodiment of this invention. It is a figure which shows the function structural example of the analyzer in embodiment of this invention. It is a figure which shows the structural example of a deterioration degree log | history memory | storage part. It is a flowchart for demonstrating an example of the process sequence which an analyzer performs.

  First, the concept in the present embodiment will be described. There are potential factors in the degradation of the quality of mobile communications, and there are various types or quality factors. Therefore, it can be considered that there is a relationship between a certain deterioration factor and the appearance (deterioration pattern) of the deterioration effect caused by the deterioration factor. When a data group for which communication quality is measured is collected, the data group is in a situation where effects of deterioration caused by various quality deterioration factors are mixed. Since the influence that appears in the measurement data due to the same deterioration factor is similar or common, grouping the measurement data based on the similarity of the influence can generate a group of deterioration patterns associated with the quality deterioration factor. Further, it is possible to estimate the deterioration factor from the deterioration pattern group.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a diagram illustrating a hardware configuration example of an analysis apparatus according to an embodiment of the present invention. The analysis apparatus 10 in FIG. 1 includes a drive device 100, an auxiliary storage device 102, a memory device 103, a CPU 104, an interface device 105, and the like that are mutually connected by a bus B.

  A program for realizing processing in the analysis apparatus 10 is provided by a recording medium 101 such as a CD-ROM. When the recording medium 101 storing the program is set in the drive device 100, the program is installed from the recording medium 101 to the auxiliary storage device 102 via the drive device 100. However, the program need not be installed from the recording medium 101 and may be downloaded from another computer via a network. The auxiliary storage device 102 stores the installed program and also stores necessary files and data.

  The memory device 103 reads the program from the auxiliary storage device 102 and stores it when there is an instruction to start the program. The CPU 104 executes a function related to the analysis device 10 according to a program stored in the memory device 103. The interface device 105 is used as an interface for connecting to a network.

  FIG. 2 is a diagram illustrating a functional configuration example of the analysis apparatus according to the embodiment of the present invention. 2, the analysis apparatus 10 includes a similarity calculation unit 11, a group generation unit 12, a group extraction unit 13, an output unit 14, and the like. Each of these units is realized by processing executed by the CPU 104 by one or more programs installed in the analysis apparatus 10. The analysis apparatus 10 also uses the deterioration degree history storage unit 15. The deterioration degree history storage unit 15 can be realized using, for example, a storage device that can be connected to the auxiliary storage device 102 or the analysis device 10 via a network.

  The degradation level history storage unit 15 stores a degradation level history (a time series vector calculated for each time width τ) for each attribute related to communication (for example, mobile communication) and related to communication quality. Remember. A plurality of attributes are acquired for one communication (for example, a request from the terminal to the server). For example, a unique identifier of a server involved in communication, identification information of a service used by communication, an ID of a base station involved in communication, a communication carrier involved in communication, an OS (Operating System) of a terminal involved in communication, etc. , An example of an attribute. The deterioration level is an index value indicating the degree (magnitude) of communication quality deterioration.

  FIG. 3 is a diagram illustrating a configuration example of the deterioration degree history storage unit. As shown in FIG. 3, the deterioration degree history storage unit 15 stores the time from each of N−1 from time 1 to time N for each attribute (for each of m attributes from attribute 1 to attribute m). The degree of deterioration is stored for the width τ.

  Note that the information stored in the deterioration degree history storage unit 15 may be collected using, for example, a technique disclosed in Japanese Patent Application Laid-Open No. 2015-115672. More specifically, the quality deterioration increment in JP-A-2015-115672 may be used as the degree of deterioration. That is, communication quality information (delay, etc.) is measured in one or more terminals (mobile terminals), and an estimated quality deterioration increment based on the quality information may be used as the deterioration degree in the present embodiment. Good.

  The similarity calculation unit 11 calculates the degree of degradation similarity for all time zones for each set of two attributes. All time zones refer to all combinations of start time and end time (ie, start time and duration). As a result, a set R having {start time, duration, attribute set, similarity} as elements is generated.

  The group generation unit 12 aggregates (groups) elements that have a common start time and duration and have a common part in a set of attributes, and generate a degradation pattern group. A degradation pattern group is a set of attributes that are affected by the same degradation factor. By generating the deterioration pattern group, the set R is updated to the set R ′. The set R ′ is a set having degradation pattern groups as elements. However, elements of the set R that do not belong to any deterioration pattern group are included in the set R ′ as they are.

  The group extraction unit 13 extracts one element from the set R ′ for each set of elements having a common attribute set. In the extraction, an element having a higher similarity and a longer duration is prioritized. Specifically, a value obtained by multiplying the similarity by a weight function based on the duration is compared between elements, and the one having the larger value is extracted.

  The output unit 14 outputs the element start time, duration, and attribute set extracted for each set of elements having a common attribute set. This set of start time, duration, and attribute is information indicating the estimation result of the influence range due to the same deterioration factor. The output form is not limited to a predetermined one. For example, display on a display device, output from a printing device, transmission to another device via a network, and the like may be performed.

  Hereinafter, a processing procedure executed by the analysis apparatus 10 will be described. FIG. 4 is a flowchart for explaining an example of a processing procedure executed by the analysis apparatus.

  In step S101, the similarity calculation unit 11 substitutes 2 for a variable i. i is a variable for storing the row number in FIG. 3 for one attribute of the attribute set for calculating the similarity. Subsequently, the similarity calculation unit 11 determines whether the value of i is equal to or less than m (S102). m is the total number of attribute types. That is, m is the number of rows in FIG.

  When i is equal to or smaller than m (Yes in S102), the similarity calculation unit 11 substitutes 1 for a variable j (S103). j is a variable for storing the row number in FIG. 3 for the other attribute of the attribute set for calculating the similarity. Subsequently, the similarity calculation unit 11 determines whether the value of j is smaller than i (S104).

  When the value of j is smaller than i (Yes in S104), the similarity calculation unit 11 substitutes 1 for the variable k (S105). k is a variable for storing the column number in FIG. 3 for the start time of the duration for calculating the similarity. Subsequently, the similarity calculation unit 11 determines whether or not the value of k is equal to or less than N−1 (S106). N is the number of slots at the start time (total number of columns in FIG. 3).

  When k is N−1 or less (Yes in S106), the similarity calculation unit 11 substitutes 2 for the variable w (S107). w is a variable for storing the duration (number of time slots). Subsequently, the similarity calculation unit 11 determines whether or not the value of w is N−k + 1 or less (S108).

  If the value of w is less than or equal to N−k + 1 (Yes in S108), the similarity calculation unit 11 calculates the similarity σ_ij (k, w) of the degradation level related to the set of the attribute a_i and the attribute a_j (S109). . The attribute a_i indicates the i-th attribute, and the attribute a_j indicates the j-th attribute. The similarity σ_ij (k, w) indicates the degree of similarity of the degree of deterioration from the start time k to the duration w with respect to the set of the attribute a_i and the attribute a_j.

Here, as an example, a method for calculating similarity using covariance will be described. However, grouping / similarity calculation may be performed using other clustering methods. In the following, t_k indicates the kth time. A_ (a_i, t_k) (i = 1,..., M, k = 1,..., N) indicates the degree of deterioration of the attribute a_i at time t_k.
σ_ij (k, w) = (1 / w) Σ (A_ (a_i, t_l) −A_ (a_i)) (A_ (a_j, t_l) −A_ (a_j))
(L = k, ..., k + w-1)
However, A_ (a_i) and A_ (a_j) are time averages of degradation degrees between time t_k and time t_ (k + w) A_ (a_i) = (1 / w) Σ (l = k,..., K + w−1) ) A_ (a_i, t_l)
A_ (a_j) = (1 / w) Σ (l = k,..., K + w−1) A_ (a_j, t_l)
Subsequently, the similarity calculation unit 11 determines whether or not the similarity σ_ij (k, w) is greater than f (w) (S110). f (w) is a function that returns a threshold value for extracting ones whose similarity σ_ij (k, w) is greater than or equal to a predetermined value. For example, f (w) = w may be set. That is, w may be a threshold value.

  When the similarity σ_ij (k, w) is larger than f (w) (Yes in S110), the similarity calculation unit 11 uses the list of (k, w, G_ij, σ_ij (k, w)) as a set R. (S111). Here, the set G_ij = {a_i, a_j} represents a set of attributes for which similarity determination has been performed. That is, the set R is a set having a list of (time IDk, duration w, attribute identifier set, similarity) as an element.

  In the case of No in step S110 or following step S111, the similarity calculation unit 11 adds 1 to w (S112), and returns to step S108. That is, steps S109 to S112 are executed for all ws equal to or less than N−k + 1.

  When w becomes larger than N−k + 1 (No in S108), the similarity calculation unit 11 adds 1 to k (S113), and returns to Step S106. That is, steps S107 to S113 are executed for all k equal to or less than N-1. When the value of k becomes the same value as N (No in S106), the similarity calculation unit 11 adds 1 to j (S114), and returns to Step S104. That is, steps S105 to S114 are executed for all j smaller than the value of i. Subsequently, when the value of j becomes the same as the value of i (No in S104), the similarity calculation unit 11 adds 1 to i (S115), and returns to step S102. That is, steps S103 to S115 are executed for all i that are equal to or less than m.

When the value of i is larger than m (No in S102), the process proceeds to step S116. At this time, for all attribute pairs (i, j = 1,..., M, i> j), all start times t_k (k = 1,..., N−1) and the number of slots w (w = 2). ,..., N−k + 1), the similarity σ_ij (k, w) is calculated, and only the similarity σ_ij (k, w) greater than f (w) is stored in the set R.
R = {(k, w, G_ij, σ_ij (k, w)) | k = 1,..., N-1, w = 2,..., N, i, j = 1,.
In step S116, the group generation unit 12 generates a set T with each set of (k, w) in the set R as an element. Subsequently, the group generation unit 12 substitutes 1 for the variable h (S117). The variable h is a variable for identifying each element in the set T.

  Subsequently, the group generation unit 12 extracts all elements including the h-th element (k, w) in the set T from the set R (S118). That is, elements with common (k, w) are extracted from the set R.

  When two or more elements are extracted (Yes in S119), the group generation unit 12 determines whether or not the complementary set of the G set of extracted elements is empty (S120). That is, it is determined whether or not there is a common part in the G group of extracted elements. When there is the common part (Yes in S120), the group generation unit 12 generates an element in which the extracted elements are integrated into one (S121). In aggregation, G is replaced with the union of the aggregated G. Further, the similarity σ is replaced by, for example, a representative value (maximum value or average value) of σ to be aggregated.

  For example, when three or more elements are extracted and a set of G of two elements has a common part, the two elements are aggregated.

  Subsequently, the group generation unit 12 adds the element generated by the aggregation to the set R ′ (S122). On the other hand, when two or more elements are not extracted in step S118 (No in S118), or when there is no common part in G of the extracted elements (No in S119), the group generation unit 12 determines in step S118. The extracted elements are added to the set R ′ as they are (S123).

  Subsequently, the group generation unit 12 adds 1 to h (S124). Steps S118 to S124 are executed for each element (that is, all (k, w) pairs) included in the set T (S125).

For example,
R = {(4, 6, G_ij, σ_ij (4, 6)), (4, 6, G_pq, σ_pq (4, 6)), ...}
In the case, the lists (4, 6, G_ij, σ_ij (4, 6)) and (4, 6, G_pq, σ_pq (4, 6)), which are sets of (k, w) = (4, 6), Extracted in step S118.
When G_ij∩G_pq = φ (empty set), no aggregation is performed.
If G_ij∩G_pq ≠ φ (empty set),
As G ′ = G_ij∪G_pq, the set R is updated to the following set R ′.
However, G ′ is a set having two or more elements. That is, since the original G (G_ij, G_pq, etc.) is a set having two elements, the number of elements becomes 3 or more when aggregated. Further, when the elements are not aggregated, the number of elements is two.
R ′ = {(4, 6, G ′, max (σ_ij (4, 6), σ_pq (4, 6)),...)
Here, an example is shown in which max (σ_ij (4,6), σ_pq (4,6)) is a new σ, but other representative values such as an average value may be adopted.

  Each element of the set R ′ is an example of the deterioration pattern group described above.

  Subsequently, as long as the number of elements in the set R ′ is larger than the number of types of G in the set R ′, that is, as long as there are elements that share G in the set R ′, the group extraction unit 13 performs step S127 and subsequent steps. Is repeated (S126).

  In step S127, the group extraction unit 13 extracts any two elements having the same G from the set R ′ (S128), where one element is (k, w, G, σ) and the other element is the other element. Are represented as (k ′, w ′, G, σ ′).

  Subsequently, the group extraction unit 13 determines whether k′−w + 1 ≦ k ≦ k ′ + w′−1 is satisfied (S128). This corresponds to a determination as to whether or not there are overlapping portions in the duration times of the two extracted elements. When k′−w + 1 ≦ k ≦ k ′ + w′−1 is satisfied (Yes in S128), the group extraction unit 13 compares σ × f (w) and σ ′ × f (w ′) (S129). ).

  When σ × f (w)> σ ′ × f (w ′) (Yes in S129), the group extraction unit 13 deletes (k ′, w ′, G, σ ′) from the set R ′ (S130). ). On the other hand, when σ × f (w) <σ ′ × f (w ′) (Yes in S131), the group generation unit 12 deletes (k, w, G, σ) from the set R ′ (S132). . When σ × f (w) = σ ′ × f (w ′) (No in S131), for example, the group generation unit 12 deletes the smaller w from the set R ′ (S133). However, the smaller σ may be deleted.

  When step S127 and subsequent steps are executed until there is no element having the same G in the set R ′ (No in S126), the output unit 14 outputs (k, w, G) of each element remaining in the set R ′. ) Is output (S134). Step S127 and subsequent steps are executed for each element that shares G. Therefore, in step S134, (k, w, G) is output for each G.

  In the output result, a highly similar deterioration effect due to a common deterioration factor continuously occurs in the time width τ × w from time t_k, and the influence range is included in the G attribute group. By reading each attribute, it is possible to specify the range of influence due to a common deterioration factor.

  In the present embodiment, the similarity calculation unit 11 is an example of a calculation unit. The group generation unit 12 is an example of an aggregation unit. The output unit 14 is an example of an output unit. The group extraction unit 13 is an example of an extraction unit.

  As mentioned above, although the Example of this invention was explained in full detail, this invention is not limited to such specific embodiment, In the range of the summary of this invention described in the claim, various deformation | transformation・ Change is possible.

DESCRIPTION OF SYMBOLS 10 Analysis apparatus 11 Similarity calculation part 12 Group generation part 13 Group extraction part 14 Output part 15 Degradation degree log | history storage part 100 Drive apparatus 101 Recording medium 102 Auxiliary storage apparatus 103 Memory apparatus 104 CPU
105 Interface device B bus

Claims (6)

With reference to the storage unit that stores the degree of deterioration of communication quality for each time width for each of a plurality of attributes related to communication in time series, all two attributes for every duration including one or more time widths A calculation unit that calculates the degree of similarity of the deterioration level for the set of
Among the calculated similarities, an aggregation unit that aggregates two or more similarities that have a common duration and at least a part of the attributes into one,
An output unit that outputs a duration related to the aggregated similarity and a set of attributes;
An analysis apparatus comprising: An extraction unit that extracts one similarity based on a similarity and a threshold value based on a duration for each similarity that has a common set of attributes among the aggregated similarities;
The output unit outputs a duration related to the extracted similarity and a set of attributes;
The analyzer according to claim 1. The calculation unit extracts a part of the similarity based on a threshold based on the duration from the calculated similarity,
The aggregating unit aggregates two or more similarities having a common duration among the extracted similarities and having at least some of the attributes in common.
The analyzer according to claim 1 or 2. With reference to the storage unit that stores the degree of deterioration of communication quality for each time width for each of a plurality of attributes related to communication in time series, all two attributes for every duration including one or more time widths A calculation procedure for calculating the similarity of the degree of deterioration for the set of
In the calculated similarity, an aggregation procedure for consolidating two or more similarities having a common duration and at least a part of the attributes into one,
An output procedure for outputting a duration related to the aggregated similarity and a set of attributes;
A computer-executed analysis method. The computer executes an extraction procedure for extracting one similarity degree based on the similarity degree and a threshold value based on the duration time for each similarity degree having a common set of attributes among the aggregated similarity degrees,
The output procedure outputs a duration related to the extracted similarity and a set of attributes.
The analysis method according to claim 4. The calculation procedure extracts a part of the similarity based on a threshold based on the duration from the calculated similarity,
The aggregation procedure aggregates two or more similarities having a common duration among the extracted similarities and having at least some of the attributes in common.
6. The analysis method according to claim 4 or 5, wherein:

Images