JP2014164382A - Information storage method, information storage device, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information storage method, an information storage device, and a program for generating a chunk data such as sensor information, sequentially storing it, and searching by relatively small computational complexity and use memory.SOLUTION: Before storing data in an information holding part 111, an information storage device 100 makes an information selection part 110 perform classification selection, classifies in advance based on the result and records information. An information extracting part 112 generates a data file called a chunk containing the data included in one of the classified storage area, and deletes the generated data from the information holding part 111.

Description

  The present invention relates to an information recording method, an information recording apparatus, and a program that constitute a sensor information database.

Conventionally, a method and an apparatus (Patent Document 1) for generating a data file called a chunk have been proposed for the purpose of efficiently storing information such as sensor measurement data, that is, sensor information.
The essential points of the basic operation of the apparatus are expressed by the following features when the description of Patent Document 1 is appropriately read in accordance with the description of the present invention described below. “An information holding unit that holds a data group including a plurality of attribute values; an information selection unit that selects information satisfying a predetermined condition with respect to the attribute value among the data held by the information holding unit; An information cutout unit that generates a file for each of the predetermined conditions including the information selected by the information selection unit, and deletes the information selected by the information selection unit from the information holding unit. Features. "
According to the above-described conventional feature, the information holding unit functions as a temporary buffer for sensor information, and sequentially generates data files called chunks that are archive files of sensor information. Chunks are generated in a state arranged according to predetermined conditions, such as “data on temperature sensors”. This condition may be determined, for example, based on the attribute value or attribute name because each data includes a plurality of attribute values. As a result, a large amount of data can be accumulated at high speed over a long period of time, the information holding unit can be realized in a certain storage area, and the chunk can be easily transferred to another computer by file transfer etc. Advantages such as being able to

  Conventionally, there has been a method for efficiently generating chunks using a tree structure algorithm such as a multidimensional search tree UBI-Tree (Non-Patent Document 2). According to this prior art, by holding the information holding unit with a tree structure algorithm, it is possible to generate chunks by collecting data that are close to each other in the tree structure, that is, data that is closely related to data values. Therefore, a flexible operation is possible even when various types of sensor information are handled at a time.

  On the other hand, in recent years, small portable devices typified by smartphones have become widespread. These devices are inferior to personal computers and server devices in terms of calculation speed, memory, etc., but they are small, light and portable, and include acceleration sensors, GPS, gyro sensors, proximity sensors, thermometers, illuminometers, and cameras. It has a feature that it is equipped with a lot of sensors such as a sound volume meter, microphone, NFC and FeliCa. In addition, it is possible to connect to the Internet via a mobile phone line, a wireless LAN, etc., and in addition to the built-in sensor, information on the Internet can be handled together. Accordingly, there is an increasing demand for sequentially storing these pieces of information.

JP 2011-170791

"High-quality chunk generation method for schemaless sensor data using multi-dimensional search tree UBI-Tree" Yutaka Arakawa et al., IEICE 2012 General Conference, B-19-16, March 2012

The chunk generation in the prior art and the information recording apparatus characterized by it have a problem that the amount of calculation (the amount of processing) and the required memory are too large when used in a smartphone or the like. Specifically, there are the following problems.
First, there is a problem that the information holding unit requires a large amount of memory and a large amount of calculation is required for the operation of selecting information from the information holding unit. For example, although one embodiment is disclosed in Patent Document 1, in this information selection operation, it is necessary to examine all data contents held by the information holding unit and select an information extraction target, and the amount of calculation is large. It was big. Further, for example, Non-Patent Document 1 discloses a form in which the information holding unit is configured by a UBI-Tree search tree, but the amount of memory occupied by this index is large.

Next, although the information cutout involves copying data from the information holding unit to the chunk, there is a problem that the amount of processing is large.
Next, there was a problem that the amount of calculation at the time of data retrieval was large. When searching for data stored in the form of chunks, if all the data is searched, a large amount of data must be investigated, and the amount of processing for this is large.

  Next, there is a problem that a memory required when an application program uses sensor information obtained by data search becomes large. When an application issues a search request to an information recording apparatus, a so-called database management system, the database management system returns search result data to the application. In an application that handles sensor information, it is required to analyze how sensor information changes for a certain period in time series, for example, when performing an acceleration analysis process, and to extract the features. For this reason, the total amount of data required is relatively large. Even if the amount of data is not a problem if analyzed on a PC, there are many cases where the memory area that can be used by application programs is limited in smartphones, and the above processing cannot be described due to memory constraints. there were.

Furthermore, it takes a certain amount of time to search for data due to the processing performance, but the new sensor information storage process must continue normally during that time, and such a request cannot be satisfied. There was a problem.
The present invention has been made in consideration of such circumstances, and its purpose is to sequentially store data such as sensor information while generating chunks and perform a search with a relatively small calculation amount and memory used. An object is to provide an information recording method, an information recording apparatus, and a program.

  The present invention has been made to solve the above-described problems. An information recording apparatus according to an embodiment of the present invention performs classification selection by an information selection unit prior to data storage in an information holding unit, and based on the result. To record information in advance. As a result, the memory amount of the information holding unit can be reduced, and the calculation amount of information cut-out can be reduced.

An information recording apparatus according to an aspect of the present invention is the above-described information recording apparatus, wherein the division is performed by creating files individually, and chunks are generated by renaming the files. This reduces the amount of information extraction processing.
In the information recording apparatus according to one aspect of the present invention, the meta information holding unit stores the range of time information included in the chunk. As a result, it is not necessary to perform access processing to a chunk data file that is not required for retrieval.

  In addition, the information recording apparatus according to an aspect of the present invention returns the results in a plurality of times when returning the search results to the application. At this time, the unit of frequency division is set to each chunk, and the chunks with the oldest generation order are processed. As a result, the application can receive the search result data in small pieces, so that the memory required for one process is reduced. Furthermore, since the application can receive the search result data in the order in which they are stored, when describing processing such as data analysis in time series order, the previously received search result can be discarded as appropriate. Less memory is required for a single process. On the other hand, an information recording apparatus for realizing this may have a simple configuration or processing, and a calculation amount and a necessary memory are small.

  An information recording apparatus according to an aspect of the present invention is the above-described information recording apparatus, and can properly handle simultaneous requests by a plurality of threads by appropriately securing and releasing exclusive control means, that is, multithread lock. . At this time, generally, during data reading (searching), writing (registration) of other data cannot be performed at the same time, and processing is often awaited. When the search result data is received in pieces, the lock is released so that a registration request for other data is received from the received application program. Furthermore, when such a data registration request is made, a new chunk may be generated due to the request, but a means for detecting such generation and an additional data search and search for the chunk. And a means for returning the result. As a result, even when data search takes time, new sensor information can be continuously accumulated, and such accumulation causes inconsistencies such as missing data in the data search results. Normal operation can be continued without any problems.

More specifically, the problem is solved as follows.
An information recording apparatus for recording a data group including a plurality of attribute values, an information selection unit for selecting and classifying data according to a predetermined condition, and an information holding for holding data in a storage area classified according to the classification result A data file called a chunk including data included in one of the divided storage areas, and the generated data is deleted from the information holding unit.

The information holding unit holds each of the divided storage areas as a file, and the information cutout unit renames the file to the file name of the chunk, thereby generating and deleting the chunk once. To do.
An information recording apparatus for recording a data group including a plurality of attribute values, the data including at least time information, an information holding unit for holding the data, and data called a chunk including a subset of the data set A file is generated, and the generated data is data that matches the conditions given to the information cutout unit to be deleted from the information holding unit, the data held by the information holding unit, and the generated chunk group And a meta information holding unit for holding the maximum value and the minimum value of the time information of the data included in the generated chunk for each chunk. If the condition includes time information, only chunks that match the time information are searched for by collating with the time information of the meta information holding unit.

  An information recording apparatus for recording a data group including a plurality of attribute values, wherein an information holding unit for holding the data and a data file called a chunk including a subset of the data set are generated, and the generated data Includes an information cut-out unit to be deleted from the information holding unit, and an information search unit for searching for data that matches the given condition for the data held by the information holding unit and the generated chunk group. The information recording apparatus further includes means for repeatedly performing data search and return of search results for all chunks in order from the chunk having the oldest generation order in the information search unit, and data search and search results for the information holding unit. The search results are returned in order from the data with the oldest registration order.

  Further, it has an exclusive control means that can operate normally even when another data registration request is received during the data search operation. The exclusive control means is used so that when a request for registration of another data is accepted at the time when the search result is returned, the processing of the registration request for that data is completed and the search processing following the search result can be continued. New chunks that did not originally exist due to the processing of registration requests for other data after adding the means to control and data retrieval and returning search results to all chunks were added during the retrieval process And a means for searching the added chunk and returning a search result.

  As described above, according to the present invention, an information recording method, an information recording apparatus, and a program for sequentially storing data such as sensor information while generating chunks and performing a search with a relatively small calculation amount and use memory are provided. Can be provided.

The block diagram which shows the apparatus structure in this Embodiment. The figure which shows the directory structure in this Embodiment. The figure which shows an example of the content of the file (203) in the directory structure shown in the said FIG. The figure which shows an example of the content of the file (206) in the directory structure shown in the said FIG. The figure which shows an example of the content of the file (204,205) in the directory structure shown in the said FIG. The figure which shows the information which the meta-information holding part shown in the said FIG. 1 hold | maintains. The figure which shows an example of the sensor data which it is going to register in this Embodiment. The flowchart of the registration process in this Embodiment. The figure which shows the content of the file (203) after a registration process, and the content of the newly produced | generated file in this Embodiment. The figure which shows the information which the meta information holding part after a registration process hold | maintains in this Embodiment. The figure which shows an example of the operation | movement sequence of a registration process and a search process in this Embodiment. The figure which shows the search conditions in this Embodiment. The flowchart of the search process in this Embodiment. The figure which shows the search result in this Embodiment. The figure which shows an example of the operation | movement sequence at the time of performing a registration process during a search process in this Embodiment. The figure which shows the example of a directory structure when the chunk is arrange | positioned in the same file system in this Embodiment.

One embodiment of the present invention will be described below.
FIG. 1 shows an apparatus configuration in the present embodiment. The information recording apparatus 100 includes a temperature sensor 103 and an acceleration sensor 104, and the program of the application 102 and the program of the library 101 operate. The library 101 includes an information selection unit 110, an information holding unit 111, an information cutout unit 112, an information search unit 113, and a meta information holding unit 114, and further has one lock 115 for protecting multithread operations. . The lock 115 is a general read-write lock (shared-exclusive lock). In addition, the information recording apparatus 100 includes a touch panel display 130, a mobile phone line data communication function 131, a wireless LAN (WiFi) communication function 132, a built-in flash memory 120, a microSD card slot 121, and the like that a general smartphone has. It comprises. It is assumed that a memory card is inserted in the card slot 121.

The application 102 calls the name “apl1”. The temperature sensor 103 and the acceleration sensor 104 are referred to as “smphone” for the major classification and “temperature” and “accel” for the minor classification, respectively.
For example, the following method can be used to construct the information recording apparatus 100 described above. First, a program for the library 101 is created in advance. Next, a program for the application 102 is created and compiled together with the created program for the library 101 to obtain an executable file for a smartphone. The executable file is installed as a smartphone application through a recording medium or a communication line. The above is an example of the construction method of the information recording apparatus 100. Such a construction method is generally performed in smartphone application development. The specific contents of the program of the application 102 and the program of the library 101 will be described in detail below.

FIG. 2 shows a directory structure. The information recording apparatus 100 mounts the internal flash memory 120 on the directory path / mnt / sdcard and mounts the microSD card on / mnt / sdcard / external_sd. Such a mounting situation is an example of a general configuration in a commercially available smartphone. FIG. 2 shows how the mounted directories and files are arranged. For example, file (203) is path / mnt / sdcard / apl1 / smphone / temperature / pool.txt
The file (204) is path / mnt / sdcard / external_sd / chunk / smphone / temperature / 01.cnk.
It is illustrated that the file is indicated by.

  The meaning of the path will be described. The directory name apl1 is derived from the name “apl1” of the application 102. The directory name “smphone” is derived from the above-mentioned general name “smphone”. The directory name temperature is derived from the above-mentioned minor classification name “temperature”. Similarly, the directory name accel is derived from the aforementioned minor classification name “accel”. The directory name chunk is derived from storing data files called chunks that are generated one after another by the operation of the apparatus described below.

  In the figure, symbol 201 indicates the information holding unit 111. In the present embodiment, the information holding unit 111 is realized by a file group arranged below a specific folder. Here, a state in which the file (203) and the file (206) are divided into two storage areas is illustrated depending on the type of sensor. That is, the file (203) is divided to hold sensor data related to the major classification “smphone” and the minor classification “temperature”, and similarly, the file (206) is classified into the major classification “smphone” and the minor classification “accel”. To hold sensor data.

Further, in the figure, symbol (207) is shown in this figure for explaining how a new file is generated by an example of the following operation. A specific generation procedure will be described in detail below.
FIG. 3 shows the contents of the file (203). As already indicated, the file holds sensor data relating to the large classification “smphone” and the small classification “temperature”, that is, information obtained from the temperature sensor 103. Here, a state in which three pieces of sensor data are held is illustrated. For example, in the row (301), the date and time (date) is December 06, 2012, 02:36:18, the major classification is smphone, the minor classification (type) is temperature, and the observation value (value) is 3.7 (degrees). ) Is one observation result data. For the sake of simplicity, hereinafter, a description such as “2012-12-06 02:36:18” is also used in the time notation. Thus, in the present embodiment, each sensor data is described in a data format composed of an arbitrary number of arrangements of “key = value”.

  Note that the sensor information referred to in the present invention is not limited to the above, and various types of information are targets. Specifically, for example, temperature, humidity, current or voltage value, fluid flow rate, substance The values measured by the sensor device including the density, brightness, noise, position, acceleration, etc. of the sensor may be handled, and the information is not limited to this, and may be information acquired via, for example, the Web or the Internet. Further, in addition to these values, information including metadata indicating sensor characteristics and states, measurement date and time, and the like may be used.

  FIG. 4 shows the contents of the file (206). As already indicated, the file holds sensor data relating to the large classification “smphone” and the small classification “accel”, that is, information obtained from the acceleration sensor 104. Here, a state in which two pieces of sensor data are held is illustrated. For example, the line (341) shows that the date (date) is 2012-12-07 13:28:10, the subject is smphone, the minor category (type) is accel, and the X-axis observation value (x ) Is 0.52, Y-axis observation (y) is 0.23, and Z-axis observation (z) is 9.78. A large value on the Z-axis indicates that gravitational acceleration is observed. The date and time of the row (342) is 2012-12-07 13:28:11, and only one second has passed from the row (341). This is because the time-axis resolution of the acceleration sensor 104 is high. Thus, the interval at which sensor data is observed and the registration operation is performed varies depending on the type of sensor.

  FIG. 5 shows the contents of the file (204) and the contents of the file (205). All the files are data of observation results related to smphone in the large classification (subject) and temperature in the small classification (type). Each file holds four sensor data. In addition, the date and time (date) are arranged in the oldest order in the file, and the file (205) with a larger file name than the file with the younger file name (204) retains the sensor data with the new date and time (date). Yes.

  In this embodiment, in this way, the observed sensor data is grouped by four to generate a data file called a chunk. The data file is stored in a separate directory for each major classification and minor classification of the sensor with a file name including a serial number. As new sensor data is observed and a registration operation is performed, such chunks are generated one after another. In the following, first, a specific processing procedure of the registration operation will be described, and then a processing procedure for performing a search operation on these sensor data will be described.

  FIG. 6 shows information held by the meta information holding unit 114. Digit (351) represents the major classification (subject), Digit (352) represents the minor classification (type), Digit (353) represents the serial number of the chunk file, Digit (354) is the minimum value of time (date) The digit (355) represents the maximum value of the time (date). On the other hand, the row (356) shows that the chunk data file whose major classification (subject) is smphone, the minor classification (type) is temperature, and the serial number of the chunk file is 01, that is, the data group included in the file (204) is the time The minimum value of (date) is 2012-12-01 05:45:12 and the maximum value is 2012-12-03 01:45:30. In fact, in the contents of the file (204) shown in FIG. 5, the line (311) is the minimum value of time and the line (314) is the maximum value of time, which matches the above description. I'm doing it.

  FIG. 7 shows sensor data to be registered. Here, it is assumed that the temperature sensor 103 obtains an observation value of 14.8 degrees at a time of 2012-12-07 13:28:20. The sensor data (401) describes the above contents. The rules of description are as described above.

FIG. 8 shows a flowchart of the registration process. A processing procedure will be described along this line.
When the application 102 calls the data registration function of the library 101, the library 101 starts registration processing (step 410). The library 101 first acquires the lock 115 (procedure 411). Since writing to data occurs here, a write lock is acquired. A method of securing a lock at the entrance of a library function and releasing it just before the end is a general multithread programming technique. Next, the information selection unit 110 acquires the classification information, that is, the values of the major classification (subject) and the minor classification (type) from the sensor data (401) (procedure 412). Here, the values “smphone” and “temperature” are obtained. Next, the library 101 stores the sensor data (401) in an area selected by the classification information among the divided storage areas in the information holding unit 111. That is, a line of sensor data (401) is added to the last line of the file (203) represented by the path names of the major classification and minor classification (procedure 413).

  Next, the information cutout unit 112 determines the size of the area, that is, the file (203), and determines whether to execute the chunk generation process (step 414). The size determination may be performed based on, for example, whether the number of lines or bytes of the file exceeds a certain threshold. In this embodiment, when the number of lines in a file becomes 4 or more, chunk generation processing is executed according to the following procedure.

The information cutout unit 112 first attempts to generate a chunk by moving a file (procedure 415). The file to be generated generates a file name including the next number in the serial number at the directory position indicated by the symbol (207) in FIG. That is, in this example,
Path / mnt / sdcard / external_sd / chunk / smphone / temperature / 03.cnk
An attempt is made to generate the file indicated by.

Here, chunk generation by moving files is
"/Mnt/sdcard/apl1/smphone/temperature/pool.txt"
A file called
"/Mnt/sdcard/external_sd/chunk/smphone/temperature/03.cnk"
This is realized by performing an operation of renaming. In general, in an OS such as Linux (registered trademark) or Windows (registered trademark), a file can be moved at the same time by specifying another directory name when renaming the file. However, the rename operation succeeds only when the source and destination are the same disk device.

  In the present embodiment, as already described, the transfer source directory exists on the built-in flash memory 120 and the transfer destination directory exists on the memory card inserted into the microSD card slot 121. That is, since the source and destination are different disk devices, the rename operation fails (success determination in step 416).

Next, the information cutout unit 112 substitutes the file move operation by copying the file and deleting the original file, thereby generating a chunk. That is, file (203)
"/Mnt/sdcard/external_sd/chunk/smphone/temperature/03.cnk"
(Step 417). Subsequently, the file (203) is deleted (step 418).

Next, the information cutout unit 112 creates a new file (203) (procedure 419). The content is empty.
FIG. 9 shows the contents of the file (203) after registration processing and the contents of a newly generated file. Lines (331) to (333) are the contents described in the original file (203). A line (334) is a line added by the procedure 413 and is the contents of the sensor data (401).

Next, the information cutout unit 112 adds information on the minimum value and the maximum value of the time included in the data file of the new chunk created by the above series of procedures to the meta information holding unit 114 (procedure 420).
FIG. 10 shows information held by the meta information holding unit 114 after the registration process. Line (359) is the information newly added in step 420, the major classification (subject) is smphone, the minor classification (type) is temperature, and the chunk with the serial number of 03 is the minimum of time (date) It represents that the value is 2012-12-06 02:36:18 and the maximum value is 2012-12-07 13:28:20. These times are the same as the time information of the row (331) and the row (334) in the generated chunk.

Finally, the library 101 releases the lock 115 (step 421), returns to the library caller, and completes a series of processing.
FIG. 11 shows an example of an operation sequence of registration processing and search processing. In FIG. 11, symbols 560 to 561 illustrate the interaction between the application 102 and the library 101 related to the registration processing described above. That is, in FIG. 11, the thread A (551) is a thread that runs in the application 102 and calls the library 101. Control is transferred to the library 101 by the data registration function call (560), and the call source is returned by return (561). Control returns to

In the following, the search process will be described next. In FIG. 11, the operations from symbol 562 to symbol 567 are performed.
FIG. 12 shows search conditions. The search condition (501) includes a time (date) range from 2012-12-04 12:00:00 to 2012-12-06 12:00:00, and a large classification (subject) of smphone and small It represents the condition of searching for data that satisfies all the conditions that the classification (type) is temperature.

  In the present embodiment, it is essential to specify the values of the major classification (subject) and the minor classification (type) as search conditions. In general, since the application 102 knows the subject and type of data registered by itself, this restriction is not a problem.

FIG. 13 shows a flowchart of search processing. The processing procedure will be described with reference to FIGS.
When the application 102 calls the data search function of the library 101 (symbol 562), the library 101 starts search processing (procedure 510). When calling, the application 102 passes a callback function as an argument in addition to the search condition (501). Such an operation can be easily realized by passing a function pointer as an argument in the programming language C, for example, and is a general programming technique. When the function is called, the information search unit 113 first acquires the lock 115 (procedure 511). Here, only reading to data occurs, so a read lock is acquired.

Next, the information search unit 113 acquires a file list of chunks as search candidates (procedure 512). The search candidate here is a chunk that matches the classification information acquired from the search condition, and may include target data against the range of time information held by the meta information holding unit 114. It is a file. Specifically, in this example, the three chunks “01.cnk”, “02.cnk”, and “03.cnk” at this point in time are classified information obtained from the search condition (501), that is, a large classification (subject). Is a storage directory corresponding to smphone and subclass (type) temperature
Path / mnt / sdcard / external_sd / chunk / smphone / temperature /
Exists. Here, for each of the three files, the corresponding line is searched from the meta information holding unit 114 to obtain the minimum value and the maximum value of the time. For example, for 01.cnk, information that the minimum value is 2012-12-01 05:45:12 and the maximum value is 2012-12-03 01:45:30 is obtained. This is compared with the time information range included in the search condition (501) to determine whether there is any overlap. For example, for 01.cnk, the range of time information included in the search condition (501) is 2012-12-04 12:00:00 to 2012-12-06 12:00:00, so there is no overlap Therefore, it can be determined that 01.cnk is not a search candidate. Similarly, for 02.cnk and 03.cnk, it is determined that there is an overlap in the range of time information, and it can be determined that these are search candidates. As a result, lists of 02.cnk and 03.cnk are obtained as the file list.

Next, the information search unit 113 releases the lock 115 (procedure 513). At this point, the library 101 can accept a data registration request from another thread in the multithread processing.
Thereafter, the file list is repeatedly executed from the step 515 to the step 517 in ascending order of the file name (step 514). In this example, the process for 02.cnk is first performed, and then the process for 03.cnk is performed.

  The information search unit 113 reads the file (procedure 515), and extracts data that matches the search condition (501) from the data in the file (procedure 516). For example, in the case of processing for 02.cnk, line (323) and line (324) are extracted. Next, when the extraction result is not empty, the information search unit 113 performs a callback using the extraction result as the search result (procedure 517 and symbol 563). The callback destination is a callback function passed from the application 102. After the process returns from the callback function (symbol 564), the information search unit 113 continues the process.

FIG. 14 shows the search result. The contents of the search result (591) passed to the callback function are as described above.
For example, in the case of processing for 03.cnk, the row (331) is extracted. This is used as a search result to make a callback (step 517 and symbol 565). The contents of the search result (592) passed to the callback function are as described above. After the process returns from the callback function (symbol 566), the information search unit 113 continues the process.

  In this way, the search result is called back multiple times, and the data included at that time is returned in the order of registration, that is, from the oldest date to the newest date. Therefore, when the application analyzes the search result data in chronological order, when the kth callback is received, the data returned by the (k-1) th callback can be discarded. As a result, it is possible to obtain the effect that the required memory amount of the application can be kept small.

  After completing processing of all files in the file list (procedure 514), the information search unit 113 acquires the lock 115 again (procedure 518). Next, the information search unit 113 acquires a file list of chunks that are search candidates and an increase from the previous acquisition (procedure 519). As described above, the increment means that the lock is released and the data registration request can be accepted during the execution of step 514 to step 517, so a new chunk may be generated with data registration. , Refers to such chunks. In order to obtain the increment, it is only necessary to store the previous acquisition result and compare it with the current file list. Further, for the file list, similarly to the procedure 512, the search candidates are narrowed down by performing a process of comparing with the time information range held by the meta information holding unit 114.

If there is an increase (procedure 520), the search processing for the increase is performed by executing procedure 513 and the subsequent steps again.
If there is no increase, the search for chunks is completed, and the remaining search candidates are only data held by the meta information holding unit 114, so the information search unit 113 matches the classification information acquired from the search conditions. Read pool.txt stored in the location (step 521). In this example, the file (203) in which the major classification (subject) corresponds to smphone and the minor classification (type) corresponds to temperature is targeted. The information search unit 113 extracts data that matches the search condition (501) from the read data (procedure 522). The information search unit 113 releases the lock 115 (procedure 523). Next, when the extraction result is not empty, the information search unit 113 performs a callback using the extraction result as the search result (procedure 524). In this example, since the result extracted in step 522 is an empty set, the callback can be omitted.

Finally, the information search unit 113 returns to the library caller (symbol 567) and completes a series of processing.
Symbol 570 indicates that it is possible to accept a data registration request from another thread A (551) during the above-described series of processing. Such a request is automatically queued by the lock 115 and processed when the lock 115 is released. The flow of processing is as described above.

  According to the above procedure, the lock 115 is released at the timing when a search process for a large number of chunks (procedure 515 to procedure 516) that is expected to take time and the callback to the application 102 is performed. Kept in a state. That is, another processing request such as a data registration request can be accepted. For this reason, for example, new sensor data can be registered even during the search process, and the occurrence of an event such as omission of registration (data missing) can be prevented. Another advantage is that new data can be registered in the callback function, which will be described in detail below.

  FIG. 15 shows an example of an operation sequence when registration processing is performed during search processing. Here, with respect to the initial state described with reference to FIGS. 1 to 6, the variant of the application (102) searches for FIG. 12, and requests registration of new data of FIG. 7 in the callback function. An example of the operation in the case of this will be described. When this is associated with the operation sequence of FIG. 11, symbol 580 corresponds to symbol 562, symbol 581 corresponds to symbol 563, symbol 582 corresponds to symbol 560, symbol 583 corresponds to symbol 561, symbol 584 Corresponds to symbol 564, symbol 585 corresponds to symbol 565, symbol 586 corresponds to symbol 566, and symbol 587 corresponds to symbol 567.

  In this operation example, first, the application 102 calls a data search function (symbol 580). Of the series of processes described above, in the candidate file list acquisition (step 512), only the chunks 01.cnk and 02.cnk exist at this time, so the file list of chunks that are search candidates is 02.cnk. Become. Then, the search process is performed on 02.cnk and the search result is called back (symbol 581). Here, in the callback function in the application 102, after calling the data registration function (symbol 582) (symbol 583), the process returns (symbol 584). At this time, since the lock 115 is in a released state, the operation of the data registration function call can be normally continued without being inhibited by the lock 115. In the data registration function call processing, a new chunk of 03.cnk is generated by the series of processing described above. After returning with the symbol 584, a file named 03.cnk newly generated as a candidate file list for the increase is acquired by the procedure 519. Since it is determined in step 520 that there is an increase, step 513 and subsequent steps are executed again, and a callback related to the search result of 03.cnk is performed (symbol 585). When returning from the callback (symbol 586), the candidate file list for the increase is acquired again by the procedure 519. Since no chunk has been generated this time, it becomes an empty list, and it is determined in step 520 that there is no increase, and the subsequent processing is performed. Through the processing described above, the application 102 can obtain the same search results as in FIG. In this way, data registration in multithread and data registration in the callback function are performed by the device of step 519 to step 520, and even if chunks are generated one after another, data search It can operate normally without inconsistency such as missing data in the result.

  As described above, in the present embodiment, by performing the search process according to the processing procedure shown in FIG. 13, the sensor information is sequentially accumulated, and search and insertion can be executed simultaneously in multi-threads, but the search results are inconsistent. Returning search results in chronological order without arriving at can be realized with a relatively simple processing procedure. I will further explain these effects.

  If it was necessary to rewrite a table like a general RDBMS instead of storing sensor information sequentially, more complicated processing would have been necessary. On the other hand, in the present embodiment, a simple processing procedure of releasing the lock 115 immediately after obtaining the chunk file list (procedure 512) (procedure 513) may be used. This is because, for the original purpose of sequentially storing sensor information, old sensor information that has been stored does not need to be rewritten, so the chunks are only increased without being rewritten. This is because the file list acquired first uses the characteristic that it is also effective during the subsequent search processing.

  If there is no need to return search results in chronological order, the processing procedure of FIG. 13 could be further simplified. That is, after acquiring the lock 115 (procedure 511) and acquiring the file list (procedure 512), the search processing related to pool.txt is performed (procedures 521 to 522), the lock is released (procedure 523), and thereafter It is only necessary to perform a callback for pool.txt (procedure 524) and search processing for chunks (procedures 514 to 517), and reconfirmation of the increase (procedures 519 to 520) is not necessary. This is because, even if data registration occurs in parallel during the search process and a new chunk is generated as a result, the file list of the chunk to be processed has already been acquired and is not affected. This is because such registration data is simply ignored and does not affect the validity of the search result. However, in such processing, search results cannot be returned in chronological order, which increases the amount of memory required for the application 102, which is a problem. On the other hand, in the present embodiment, the search results are returned in chronological order, and inconsistencies that may be caused by the search results are detected in steps 519 to 520, and the operation for re-execution of processing is newly incorporated. The advantage is obtained that the requirements can be met with a relatively simple procedure.

  Needless to say, if there was no need for multi-thread processing, these contradictions could not occur in the first place, and therefore the processing procedure of FIG. 13 could be greatly simplified. However, in such processing, new data registration cannot be performed in parallel during the search processing, which hinders the original purpose of the application 102. On the other hand, in this embodiment, there is an advantage that new data can be registered in parallel from another thread and callback function.

  As described above, in the present embodiment, chunks are stored in different directories for each type of sensor. This also provides the advantage that old chunks can be deleted at different frequencies as needed for each sensor type. That is, for example, in the present embodiment, the temperature sensor 103 has an observation period of about once or twice per day, and the acceleration sensor 104 has an observation period of about once per second. For this reason, considering the free capacity of the storage device, there is a request that the data of the temperature sensor 103 is stored for about one year while the data of the acceleration sensor 104 is to be discarded in about one hour. In this embodiment, since these two data are stored separately in different directories, for example, only a directory where the chunks of the acceleration sensor 104 are stored is monitored at regular intervals, and a file that has passed one hour or more is The advantage that it is easy to perform the process of deleting is obtained.

  In the series of processes described above, as shown in FIG. 2, the directory structure is such that chunks are generated on the microSD card. Generally, the built-in flash memory 120 has a small capacity but a high speed, and an external memory card tends to have a large capacity but a low speed. In the present embodiment, the information holding unit 111 that repeats writing frequently but does not require a certain storage area is arranged in the built-in flash memory 120, and the chunk is generated on the memory card inserted in the microSD card slot 121. By doing so, a feature that can store a large amount of data can be obtained. On the other hand, a configuration in which chunks are generated on the built-in flash memory 120 is also possible, and the operation in that case will be described below.

FIG. 16 shows a directory configuration example when chunks are arranged in the same file system. This configuration example is different in that the directory (202) in FIG. 2 is arranged in the directory (210) in FIG.
When the data registration process shown in FIG. 8 is performed with this directory structure, the file is moved to the chunk by renaming (procedure 415) because the source and destination are the same file system. Therefore, step 417 to step 418 can be omitted. That is, there is an advantage that the file copy and the old file delete operation can be completed by a single rename operation. As a result, it is possible to obtain an effect of reducing the amount of information extraction processing from the information holding unit 111 to the chunk.

The apparatus of the present invention can also be realized by a computer and a program, and the program can be recorded on a recording medium or provided through a network.
In short, the present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. Further, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, you may combine suitably the component covering different embodiment.

  DESCRIPTION OF SYMBOLS 100 ... Information recording apparatus, 101 ... Library, 102 ... Application, 103 ... Temperature sensor, 104 ... Acceleration sensor, 110 ... Information selection part, 111 ... Information holding part, 112 ... Information extraction part, 113 ... Information search part, 114 ... Meta information holding unit, 115 ... lock, 120 ... built-in flash memory, 121 ... microSD card slot, 130 ... touch panel display, 131 ... data communication function, 132 ... LAN (WiFi) communication function.

Claims (11)

An information recording method for recording a data group including a plurality of attribute values,
A first procedure for classifying and selecting data according to predetermined conditions;
A second procedure for holding data in the divided storage area of the information holding unit according to the classification result;
And a third procedure for generating a data file called a chunk including data included in one of the divided storage areas and deleting the generated data from the information holding unit. Information recording method. The second procedure holds each of the divided storage areas as a file,
The information recording method according to claim 1, wherein the third procedure performs generation and deletion of the chunk at a time by renaming the file to a file name of the chunk. An information recording method for recording a data group including a plurality of attribute values,
The data contains at least time information,
A first procedure for holding the data in the information holding unit;
Generating a data file called a chunk containing a subset of the data set, and deleting the generated data from the information holding unit;
A third procedure for searching for data that matches a given condition for the data held by the information holding unit and the generated chunk group;
A fourth procedure for holding a maximum value and a minimum value of time information of data included in the generated chunk for each chunk, and
In the third procedure, when the given condition includes time information, only chunks matching the time information are searched by collating with the time information of the fourth procedure. Information recording method. An information recording method used in an information recording apparatus for recording a data group including a plurality of attribute values,
A first procedure for holding data in the information holding unit;
Generating a data file called a chunk containing a subset of the data set, and deleting the generated data from the information holding unit;
A third procedure for searching for data that matches a given condition for the data held by the information holding unit and the generated chunk group, and
In the third procedure,
Processing to repeat data search and return search results for all chunks in order from the chunk with the oldest generation order,
An information recording method characterized in that the search results are returned in order from the data with the oldest registration order by sequentially executing a data search for the information holding unit and a process of returning the search results. Furthermore, the information recording apparatus has an exclusive control means that can operate normally even when a registration request for another data is received during a data search operation,
In the third procedure, when the search results are sequentially returned, when the request for registration of another data is accepted at the time when one search result is returned, the processing of the registration request for the data is completed and the search result is returned. A procedure for controlling using the exclusive control means so that the search processing subsequent to can be continued,
Check that new chunks that were not originally present due to the processing of other data registration requests after the data search and return of search results are repeated for all chunks are added during the search process. Steps to detect,
The information recording method according to claim 4, further comprising a procedure for performing a data search and a return of a search result for the added chunk. An information recording apparatus for recording a data group including a plurality of attribute values,
An information selection unit that classifies and selects data according to predetermined conditions;
An information holding unit for holding data in the storage area sorted according to the classification result;
A data file called a chunk including data included in one of the divided storage areas is generated, and the generated data includes an information cutout unit that is deleted from the information holding unit. Information recording device. The information holding unit holds each of the divided storage areas as a file,
7. The information recording apparatus according to claim 6, wherein the information cutout unit performs generation and deletion of the chunk at a time by renaming the file to a file name of the chunk. An information recording apparatus for recording a data group including a plurality of attribute values,
The data includes at least time information,
An information holding unit for holding data;
Generating a data file called a chunk including a subset of the data set, the generated data being deleted from the information holding unit;
An information search unit that searches for data held by the information holding unit and data that matches a given condition for the generated chunk group; and
A meta information holding unit that holds a maximum value and a minimum value of time information of data included in the generated chunk for each chunk;
When the given condition includes time information, the information search unit searches only chunks that match the time information by collating with the time information of the meta information holding unit. Recording device. An information recording apparatus for recording a data group including a plurality of attribute values,
An information holding unit for holding data;
Generating a data file called a chunk including a subset of the data set, the generated data being deleted from the information holding unit;
An information search unit that searches for data that matches the given condition for the data held by the information holding unit and the generated chunk group, and
The information recording apparatus further includes:
A means of repeatedly performing data search and returning of search results for all chunks in order from the chunk with the oldest generation order,
An information recording apparatus characterized in that the search results are returned in order from the data with the oldest registration order by sequentially executing a data search for the information holding unit and a means for returning the search results. Furthermore, it has an exclusive control means that can operate normally even when another data registration request is received during the data search operation,
When the search result is sequentially returned, the information search unit completes the processing of the registration request of the data when the request for registration of another data is accepted at the time when the search result is returned, and the search result is displayed. Means for controlling using the exclusive control means so that subsequent search processing can be continuously executed;
Check that new chunks that were not originally present due to the processing of other data registration requests after the data search and return of search results are repeated for all chunks are added during the search process. Means for detecting;
10. The information recording apparatus according to claim 9, further comprising means for performing a data search and returning a search result for the added chunk. A program executed by an information recording apparatus for recording a data group including a plurality of attribute values, the information recording apparatus comprising:
Select and classify data according to predetermined conditions,
According to the classification result, data is held in the storage area divided by the information holding unit,
A program that generates a data file called a chunk including data included in one of the divided storage areas, and operates to delete the generated data from the information holding unit.

Images