JP5737718B2 - Apparatus, program, and method for estimating life pattern of user having portable terminal - Google Patents

Description

  The present invention relates to a technique for estimating a position associated with movement of a user carrying a mobile terminal.

  In recent years, a mobile terminal typified by a mobile phone has generally been equipped with a positioning function such as GPS (Global Positioning System). Therefore, the user can measure the current position using the mobile terminal, and can receive various service information corresponding to the position by transmitting the position to the server via the network.

  Conventionally, there is a technology for transmitting position information acquired by the GPS function of a mobile terminal to a server, and that the server calculates an action range from the user's action history and provides information reflecting the action range (for example, Patent Document 1). According to this technique, clustering is performed based on the distance between position information measured by the mobile terminal.

  According to such technology, while it is possible to calculate the action range of the user who possesses the mobile terminal, it also recognizes what life pattern the user has (business, private, everyday or extraordinary) I can't do it. Even when the server provides service information corresponding to the position and time to the mobile terminal, the service information to be provided differs depending on the lifestyle pattern corresponding to the position and time. For example, the content of service information to be provided is completely different between the case of taking a daily life pattern and the case of taking an extraordinary life pattern.

  On the other hand, there is a technique for learning a significant position of a user from an action history based on position information acquired by a GPS function of a mobile terminal (see, for example, Non-Patent Document 1). According to this technique, the position information needs to be periodically acquired by the GPS function.

  There is also a technique for estimating an action pattern frequently performed by a user from an action history based on a base station used for communication (for example, see Non-Patent Document 2).

JP 2010-49295 A

Toyama Midori, Hattori Takashi, Kanno Tatsuya, “Learning Significant Positions Using Cellular Phone Positioning Functions”, IPSJ Transactions, vol.46 No.12, pp.2915-2924, 2005 MA Bayir, M. Demirbas, and N. Eagle, `` Mobility profiler: Aframework for discovering mobility profiles of cell phone users, '' Proc. Of the International Conference on Pervasive and Mobile Computing, vol.6, no.4, pp. 435--454, 2010

  According to the above-described conventional technology, it is assumed that the position information of the user is acquired by the GPS function of the mobile terminal. However, for a mobile terminal possessed by a user, starting the GPS function and its application constantly or periodically not only accelerates the battery consumption of the mobile terminal, but also increases the amount of packets sent from the mobile terminal. There was a problem. For this reason, it is difficult to widely disseminate services based on the user's home range.

  Further, according to the technique described in Non-Patent Document 1, it is possible to estimate the stay position when the measurement information interval of the position information is substantially constant. However, when the measurement time interval of the position information is not constant, it is difficult to estimate the stay position, and it is further difficult to estimate the life pattern of the user.

  Furthermore, according to the technique described in Non-Patent Document 2, base stations used for communication can be clustered, and a user's action pattern can be estimated based on a time-series combination pattern of clusters. However, when the communication interval is intermittent, the same life pattern may be recognized as different patterns due to the effect of the missing information, and there is a problem that the life pattern cannot be sufficiently recognized.

  In contrast to such conventional technology, the telecommunications carrier side can estimate the significant position of the mobile terminal not based on the position information acquired by the mobile terminal but based on the position information of the base station to which the mobile terminal is subordinate. preferable. In this case, it is not necessary for the portable terminal to activate the GPS function constantly or periodically.

  Therefore, in the present invention, without starting the positioning function of the mobile terminal, using the history of base station location information that can be acquired by the telecommunications carrier equipment, the spatial granularity is coarse and the time interval is not constant, An object of the present invention is to provide an apparatus, a program, and a method capable of estimating a precise life pattern by appropriately interpolating the spatiotemporal information.

According to the present invention, a life pattern estimation device connected to a wide area wireless communication network, which can estimate a life pattern of a user who has a mobile terminal,
For a base station provided with a mobile terminal, a base station location information registration unit that registers and registers location information for each base station identifier;
Life pattern template registration means for registering a plurality of life pattern templates associated with life pattern identifiers for each time period;
From the base station, communication history information collecting means for collecting a communication history including a lot of data consisting of date and time and base station identifier for each portable terminal,
Base station position information adding means for adding position information corresponding to the base station identifier to each data of the communication history using the base station position information registering means;
Time division means for dividing communication history data according to each time zone for each life pattern template;
Spatio-temporal interpolation means for temporally and spatially interpolating communication history data divided according to each time zone for each life pattern template ;
And a spatio-temporal interpolation result evaluating means for evaluating the interpolation result of the spatiotemporal interpolation means with reference to an accumulated error.

  According to another embodiment of the life pattern estimation apparatus of the present invention, the spatiotemporal interpolation means converts the spatiotemporal information of the first day to the time of the second day based on the position information generated regularly every day. It is also preferable to interpolate the spatiotemporal information of different days so that it is supplemented with the spatial information, and conversely, the spatiotemporal information of the second day is supplemented with the spatiotemporal information of the first day.

  According to another embodiment of the life pattern estimation apparatus of the present invention, the life pattern template showing the best result and the evaluation result of the spatiotemporal interpolation based thereon are registered as the evaluation result of the spatiotemporal interpolation result evaluation means. It is also preferable to further include a life pattern estimation / registration means.

  According to another embodiment of the life pattern estimation apparatus of the present invention, it is preferable that the spatiotemporal interpolation means linearly transforms x and y as pre-processing.

  According to another embodiment of the life pattern estimation apparatus of the present invention, it is also preferable that the spatiotemporal interpolation means optimizes the interpolation parameter using an EP method (expected value propagation method) or a variational Bayes method.

According to another embodiment of the life pattern estimation apparatus of the present invention,
The life pattern template registration means estimates the stay position in the time zone by clustering base station identifiers based on the communication history of the mobile terminal, and the movement process from the first stay place to the second stay place in one life If the combination of the departure place and the arrival place is the same, it is also preferable to use the same life pattern.

According to the present invention, a program for causing a computer mounted on a device connected to a wide area wireless communication network to function and estimating a life pattern of a user possessing a mobile terminal,
For a base station provided with a mobile terminal, a base station location information registration unit that registers and registers location information for each base station identifier;
Life pattern template registration means for registering a plurality of life pattern templates associated with life pattern identifiers for each time period;
From the base station, communication history information collecting means for collecting a communication history including a lot of data consisting of date and time and base station identifier for each portable terminal,
Base station position information adding means for adding position information corresponding to the base station identifier to each data of the communication history using the base station position information registering means;
Time division means for dividing communication history data according to each time zone for each life pattern template;
Spatio-temporal interpolation means for temporally and spatially interpolating communication history data divided according to each time zone for each life pattern template ;
The computer is caused to function as a spatiotemporal interpolation result evaluation unit that evaluates an interpolation result of the spatiotemporal interpolation unit with reference to an accumulated error.

According to the present invention, a life pattern estimation method using a device connected to a wide area wireless communication network, which can estimate a life pattern of a user who has a mobile terminal,
The device is
A base station location information registration unit that registers and associates location information for each base station identifier with respect to a base station that is provided with a mobile terminal.
A life pattern template registration unit that registers a plurality of life pattern templates in which life pattern identifiers are associated with each time period;
A first step of collecting a communication history including a lot of data consisting of a date and time and a base station identifier for each portable terminal from the base station;
A second step of adding position information corresponding to the base station identifier to each data of the communication history using the base station position information registration means;
A third step of dividing the communication history data according to each time zone for each life pattern template;
A fourth step of interpolating temporally and spatially the communication history data divided according to each time zone for each life pattern template ;
And a fifth step of evaluating the interpolation result of the spatiotemporal interpolation means on the basis of the accumulated error.

  According to the apparatus, program, and method of the present invention, the history of base station location information that has a coarse spatial granularity and has a non-constant time interval, which can be acquired by a telecommunications carrier facility without activating the positioning function of the mobile terminal. It can be used to estimate precise life patterns.

It is a spatial external view showing movement of a portable terminal. It is a functional block diagram of the life pattern estimation apparatus in this invention. It is a graph of the result of having applied the space-time interpolation process to each data with respect to the latitude data of Table 7 for every life pattern. It is a graph of the result of having applied the space-time interpolation process to each data with respect to the longitude data of Table 8 for every life pattern. FIG. 5 is a graph of xy coordinates obtained by combining the graphs of FIGS. 3 and 4. It is explanatory drawing showing the life pattern for every day of the week estimated by the life pattern estimation and registration part.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a spatial external view showing movement of a mobile terminal.

  A mobile terminal (for example, a mobile phone) possessed by a user is always under the control of a base station and continues to receive radio waves from the base station, regardless of the position. According to FIG. 1, the user has a home address in “Fujimino City, Saitama Prefecture” and a work place in “Iidabashi, Minato-ku, Tokyo”. The user commute between home and work via Ikebukuro. Also, the user frequently moves to “Otemachi, Chiyoda-ku, Tokyo” as a visit destination.

  In a carrier equipment that integrates a large number of base stations, base station position information with a coarse spatial granularity and a constant time interval can always be collected for each mobile terminal. “Coarse spatial granularity” means that the geographical distance between location information is relatively long. Further, “the time interval is not constant” means that the acquisition time intervals of the position information are relatively varied.

  FIG. 2 is a functional configuration diagram of the life pattern estimation apparatus according to the present invention.

The base station 3 connected to the wide area wireless communication network (mobile phone network) recognizes the mobile terminal (mobile phone) 2 located under the base station 3 by exchanging radio frame signals, and acquires the date and time. Then, the base station 3 sends the data including the “terminal identifier” (address, telephone number, identification number, etc.), “base station identifier”, and “date / time” of the mobile phone 2 to the life pattern estimation apparatus 1. Send.
Data (terminal identifier, base station identifier, date and time)

  The life pattern estimation device 1 is connected to a wide area wireless communication network (a mobile phone network) and estimates a life pattern of a user who has a mobile terminal (a mobile phone). According to FIG. 2, the life pattern estimation apparatus 1 includes a communication interface unit 10 connected to a communication carrier network, a communication history information collection unit 11, a base station position information addition unit 12, a time division unit 13, a time division unit, It includes a space interpolation unit 14, a spatio-temporal interpolation result evaluation unit 15, a life pattern estimation / registration unit 16, an application processing unit 17, a base station position information registration unit 18, and a life pattern template registration unit 19. The application processing unit 17 executes various services based on the life patterns for each user estimated by the present invention. These functional components excluding the communication interface unit are realized by executing a program that causes a computer installed in the apparatus to function.

[Communication history information collection part]
The communication history information collection unit 11 collects a communication history including a lot of data including date and time and base station identifier for each mobile terminal from the base station.

Table 1 shows communication history data collected for one mobile terminal.

"Base Station Location Information Registration Department"
The base station location information registration unit 18 registers location information for each base station identifier in association with the base station provided with the mobile terminal. Table 2 shows the position information (latitude / longitude) of the base station managed by the base station position information registration unit 18.

[Base station location information addition unit]
The base station location information adding unit 12 uses the base station location information registration unit 18 to add location information corresponding to the base station identifier to each data of the communication history. Table 3 shows information generated by the base station position information adding unit 12. According to Table 3, the position information in Table 2 is associated with the base station identifier of each data in Table 1.

[Life pattern template registration section]
The life pattern template registration unit 19 registers a plurality of life pattern templates each associated with a life pattern identifier for each time slot (time range between times). Table 4 represents one template.

The “life pattern” has a specific meaning such as business or private (however, it has only a signified meaning here). For example, various life patterns can be assumed as follows.
・ Give life patterns to the days of the week (time zones)
・ Give life patterns to the time of day.
-In cooperation with the schedule book, a life pattern is given from the time zone and schedule correspondence table described in the calendar / schedule book.
By clustering base station identifiers based on the communication history of mobile terminals, it is possible to estimate the stay position (stay place) in the time zone. And the movement process from the 1st place of stay to the 2nd place of stay can be made into one life pattern. That is, when the combination of the departure place and the arrival place is the same, the same life pattern is set.

Table 5 shows a life pattern template created from a time zone and a stay place by deriving a stay place by clustering base station identifiers.

[Time division part]
The time division unit 13 divides the communication history data according to each time zone for each life pattern template.

Table 6 shows the data of the communication history shown in Table 3 to which the life pattern identifier according to each time zone for each life pattern template shown in Table 4 is given.

[Space-time interpolation unit]
The spatiotemporal interpolation unit 14 interpolates temporally and spatially the communication history data divided for each life pattern.

The following table is divided and converted as pre-processing for spatial interpolation.
Table 7: Data on “Date / Longitude Combination” in Table 6 Table 8: Data on “Date / Longitude Combination” in Table 6 Data

(1) The date / time data is converted to divide the number of seconds elapsed from the reference date / time by the number of seconds per day (86400). In Tables 7 and 8, the reference date is 2010/6/19 (Sunday) 00:00, and the time t is obtained by dividing the number of seconds elapsed from that date by 86400. As a result of this conversion, the time difference after the conversion becomes 1 for the time of the day.
(2) A vector having longitude and latitude elements (longitude and latitude) is linearly converted, and a converted vector (x, y) is calculated. Here, it is assumed that x and y are obtained by conversion by subtracting the average for each of longitude and latitude and dividing by the standard deviation. Instead of this conversion, x and y may be calculated by polar coordinate conversion or singular value decomposition.

FIG. 3 is a graph showing the result of applying the spatio-temporal interpolation process to each data for each life pattern with respect to the latitude data in Table 7.
FIG. 4 is a graph showing the result of applying the spatio-temporal interpolation process to each data for each life pattern with respect to the longitude data in Table 8.
Marker: “+”: data of (x, y) Thin curve: interpolation result based on data of “life pattern = 1” Dark curve: interpolation result based on data of “life pattern = 2”

  FIG. 5 is a graph of xy coordinates obtained by combining the graphs of FIGS. 3 and 4.

The spatio-temporal interpolation process is executed by the following two arithmetic processes. The following calculation process is used to supplement the first day's spatiotemporal information with the second day's spatiotemporal information based on the position information that occurs regularly every day. It is possible to interpolate the spatiotemporal information of different days so that the information is supplemented with the spatiotemporal information of the first day.
[Calculation 1] Calculate covariance matrix [Calculation 2] Calculate predicted mean values x *, y * and predicted variance values var (x *), var (y *) at time t * to be interpolated. The case of the interpolation process based on the data of “life pattern = 1” will be described. In particular, time position information not included in the history is interpolated using statistical information regarding position information in the same time zone on another day included in the history.

[Calculation 1] Calculation of Covariance Matrix The covariance between times t and t ′ is calculated by substituting into the following function.
||: Operation that takes an absolute value σ _f ² : Maximum variance of objective variable ω: Period l: Smoothing parameter σ _f ² = 1, ω = 1, and l = 1, t = 0.559, t ′ = 0.756 The covariance between is:

These are calculated for all combinations of times t and t ′, and the following matrix K is obtained.

Also, the covariance between time t and time t * to be interpolated is calculated in the same manner. Here, assuming t = 0.559 and t * = 0.625, the following results.

These are calculated for all combinations of covariance between time t and time t * to be interpolated to obtain the following matrix K *.

Further, the variance at time t * to be interpolated is represented as K **. In this case, since σ _f ² , K ** = 1.00.

[Calculation 2] Calculate predicted average values x *, y * and predicted variance values var (x *) and var (y *) at time t * to be interpolated using the covariance matrix calculated in [Calculation 1] Calculate with the following formula.
Bold character x: column vector of x data of “life pattern = 1” Bold character y: column vector of y data of “life pattern = 1”

For example, for t * = 2.5, the following result is obtained.

Between the processes of [Calculation 1] and [Calculation 2], parameters such as σ _f ² , ω, and l may be optimized using the EP method (expected value propagation method) or the variational Bayes method. Time t * to be interpolated is assumed to be in increments of 1 hour, and [Calculation 2] is executed for the number of increments.

Table 11 below shows the results of executing [Calculation 1] and [Calculation 2] with t * in increments of 1 hour. In addition, FIG. 5 displays the result of the following Table 11 as a graph.

Table 12 below is obtained by reversely converting the preparatory processing for space interpolation with respect to Table 11. By reverse conversion, it can be interpreted as date, latitude, and longitude.

  [Calculation 1] and [Calculation 2] described above are executed for each life pattern template by the number of life patterns included in the template.

[Time-space interpolation result evaluation section]
The spatiotemporal interpolation result evaluation unit 15 evaluates the interpolation result of the spatiotemporal interpolation unit 14 based on the accumulated error. Based on the result of executing the spatio-temporal interpolation processing, for each data, calculate the sum of squares of errors with the interpolation results for each life pattern of the life pattern templates, and obtain the total sum of squares of errors for each life pattern template. , The matching degree of the life pattern template.

As an evaluation procedure, the following two processes are executed.
[Process 1] The x and y data are interpolated using the set of time t * to be interpolated as the same set of time t as the data used in the interpolation process.

[Process 2] The sum of squared errors is calculated.
The sum of error squares of x is 0.141, the sum of error squares of y is 0.067, and the sum is 0.208.

Table 14 shows an example of the matching degree of each life pattern template. The matching degree of life pattern template = 1 is the highest. Therefore, “life pattern template = 1” is set as an optimal life pattern template.

[Life Pattern Estimation / Registration Department]
The life pattern estimation / registration unit 16 sets the life pattern template that is determined to have the highest degree of match as the result of the spatiotemporal interpolation result evaluation as the optimum life pattern template. The spatial and temporal interpolation results based on the optimal life pattern template are combined and registered as a user life pattern.

Table 15 shows an example of registered data. Table 15 gives life patterns to Table 13.

  FIG. 6 is an explanatory diagram showing a daily life pattern estimated by the daily life pattern estimation / registration unit.

[Application section]
The application unit 17 executes various services based on the life pattern for each user registered by the life pattern estimation / registration unit 16 and the spatiotemporal interpolation result.

  As described above in detail, according to the apparatus, program, and method of the present invention, the spatial granularity and time interval that can be acquired by the telecommunications carrier equipment without activating the positioning function of the mobile terminal are large. A precise life pattern can be estimated using a history of non-constant base station position information.

  According to the present invention, since it is not necessary for the mobile terminal to activate a positioning function such as GPS, it is not necessary to consider the battery consumption of the mobile terminal. That is, all information for estimating the user's life pattern can be acquired only by the telecommunications carrier. As a service that can be provided by estimating a user's life pattern, for example, there is a personalized information providing service that distributes coupon information or the like corresponding to the life pattern for each user.

  Various changes, modifications, and omissions of the above-described various embodiments of the present invention can be easily made by those skilled in the art. The above description is merely an example, and is not intended to be restrictive. The invention is limited only as defined in the following claims and the equivalents thereto.

DESCRIPTION OF SYMBOLS 1 Life pattern estimation apparatus 10 Communication interface part 11 Communication history information collection part 12 Base station position information addition part 13 Time division part 14 Space-time interpolation part 15 Space-time interpolation result evaluation part 16 Life pattern estimation and registration part 17 Application processing part 18 Base station location information registration unit 19 Living pattern template registration unit 2 Mobile terminal, mobile phone 3 Base station

Claims (8)

A life pattern estimation device connected to a wide area wireless communication network, which can estimate a life pattern of a user possessing a mobile terminal,
For a base station provided with a mobile terminal, a base station location information registration unit that registers and registers location information for each base station identifier;
Life pattern template registration means for registering a plurality of life pattern templates associated with life pattern identifiers for each time period;
From the base station, communication history information collecting means for collecting a communication history including a lot of data consisting of a date and time and a base station identifier for each portable terminal;
Base station position information adding means for adding position information corresponding to the base station identifier to each data of the communication history using the base station position information registering means;
Time division means for dividing the communication history data according to each time zone for each life pattern template;
Spatio-temporal interpolation means for interpolating temporally and spatially with respect to communication history data divided according to each time zone for each life pattern template ;
A life pattern estimation apparatus comprising: a spatiotemporal interpolation result evaluation means for evaluating an interpolation result of the spatiotemporal interpolation means on the basis of an accumulated error.   The spatio-temporal interpolation means supplements the spatio-temporal information of the first day with the spatio-temporal information of the second day based on the positional information that occurs regularly every day, and conversely, The life pattern estimation apparatus according to claim 1, wherein the spatial and temporal information of different days is interpolated so as to supplement the spatial and spatial information with the temporal and spatial information of the first day.   The life pattern estimation / registration means for registering the life pattern template showing the best result and the evaluation result of the spatiotemporal interpolation based thereon as the evaluation result of the spatiotemporal interpolation result evaluation means, Item 3. A lifestyle pattern estimation apparatus according to item 1 or 2.   4. The life pattern estimation apparatus according to claim 1, wherein the spatiotemporal interpolation unit linearly converts x and y as pre-processing. 5.   5. The life pattern estimation according to claim 1, wherein the spatiotemporal interpolation means optimizes an interpolation parameter by using an EP method (expected value propagation method) or a variational Bayes method. apparatus.   The lifestyle pattern template registration means estimates the stay position in the time zone by clustering base station identifiers based on the communication history of the mobile terminal, and moves the process from the first stay place to the second stay place as one The life pattern estimation device according to claim 1, wherein the life pattern is the same when the combination of the departure place and the arrival place is the same. A program for estimating a lifestyle pattern of a user who has a mobile terminal by causing a computer mounted on a device connected to a wide area wireless communication network to function.
For a base station provided with the mobile terminal, a base station position information registration unit that registers and registers position information for each base station identifier;
Life pattern template registration means for registering a plurality of life pattern templates associated with life pattern identifiers for each time period;
From the base station, communication history information collecting means for collecting a communication history including a lot of data consisting of a date and time and a base station identifier for each portable terminal;
Base station position information adding means for adding position information corresponding to the base station identifier to each data of the communication history using the base station position information registering means;
Time division means for dividing the communication history data according to each time zone for each life pattern template;
Spatio-temporal interpolation means for interpolating temporally and spatially with respect to communication history data divided according to each time zone for each life pattern template ;
A life pattern estimation program for causing a computer to function as a spatiotemporal interpolation result evaluation means for evaluating an interpolation result of the spatiotemporal interpolation means on the basis of an accumulated error. A life pattern estimation method using a device connected to a wide area wireless communication network, which can estimate a life pattern of a user who has a mobile terminal,
The device is
For a base station provided with the mobile terminal, a base station location information registration unit that registers and associates location information for each base station identifier,
A life pattern template registration unit that registers a plurality of life pattern templates in which life pattern identifiers are associated with each time period;
A first step of collecting a communication history including a lot of data consisting of a date and time and a base station identifier for each portable terminal from the base station;
A second step of adding location information corresponding to the base station identifier to each data of the communication history using the base station location information registration means;
A third step of dividing the communication history data according to each time zone for each life pattern template;
A fourth step of interpolating temporally and spatially the communication history data divided according to each time zone for each life pattern template ;
A lifestyle pattern estimation method comprising: a fifth step of evaluating an interpolation result of the spatiotemporal interpolation means on the basis of an accumulated error.